diff --git a/.DS_Store b/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..a87e3954609fae2a4b9bcab1ab6d7a527d91d943
Binary files /dev/null and b/.DS_Store differ
diff --git a/adapter_config.json b/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/adapter_model.safetensors b/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d4507be88293606900a542574f09ce87322b185b
--- /dev/null
+++ b/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e566c61bed3b05cde1ebb3a325febfce0fafe3d3f3e648acb48bf8bd33cedb20
+size 5919456
diff --git a/checkpoints/.DS_Store b/checkpoints/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..319da2ad88eb2d840c5fe704eb9e32c055d29681
Binary files /dev/null and b/checkpoints/.DS_Store differ
diff --git a/checkpoints/checkpoint-1000/.DS_Store b/checkpoints/checkpoint-1000/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..5008ddfcf53c02e82d7eee2e57c38e5672ef89f6
Binary files /dev/null and b/checkpoints/checkpoint-1000/.DS_Store differ
diff --git a/checkpoints/checkpoint-1000/adapter_config.json b/checkpoints/checkpoint-1000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-1000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-1000/adapter_model.safetensors b/checkpoints/checkpoint-1000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ada62a9056d2bc3e2d2cbb277e551f55e6b4aa0b
--- /dev/null
+++ b/checkpoints/checkpoint-1000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7021d0f53b2af54bc0fb2b36f002603368f00cc5e37cf8ffa4d35e14e850cfc
+size 5919456
diff --git a/checkpoints/checkpoint-1000/optimizer.pt b/checkpoints/checkpoint-1000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e68730afa32e5f46d317ce668b5fa018beb68976
--- /dev/null
+++ b/checkpoints/checkpoint-1000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cae149eaee16a95e002bf921dbb3bd8869b631f62813911b3680edf18d7ff723
+size 11930938
diff --git a/checkpoints/checkpoint-1000/rng_state_0.pth b/checkpoints/checkpoint-1000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..63c03aea1b9f7d3ab56583f85946ad1c03bba717
--- /dev/null
+++ b/checkpoints/checkpoint-1000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:287014c23e0c4e9613e974e969516cd0fc0bdce46daf1c6f1c0b66e3eb091e50
+size 15024
diff --git a/checkpoints/checkpoint-1000/rng_state_1.pth b/checkpoints/checkpoint-1000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fced8d099bdfbcd4544fef78e3314059f786c5cd
--- /dev/null
+++ b/checkpoints/checkpoint-1000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ab87bd92d10c436f5b79e618401a8481e12e99083829692c2fc0de84edcad99
+size 15024
diff --git a/checkpoints/checkpoint-1000/rng_state_2.pth b/checkpoints/checkpoint-1000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..10c8b81e903b01c272edbdb09a6d11cdd8111bb9
--- /dev/null
+++ b/checkpoints/checkpoint-1000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8553566dff78f7c07a5a3e5517a0a640c6ef80a0e95eb328fed4c566945f6fd0
+size 15024
diff --git a/checkpoints/checkpoint-1000/rng_state_3.pth b/checkpoints/checkpoint-1000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0cc13eb6a1f6df01cb785380fb1717d312f38f54
--- /dev/null
+++ b/checkpoints/checkpoint-1000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60c899ef713ad40446331266b18149afa90e55eed1210839243a923ea8aa772d
+size 15024
diff --git a/checkpoints/checkpoint-1000/scheduler.pt b/checkpoints/checkpoint-1000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a0d6f7f9da279b71ea03802b24c083ad94591354
--- /dev/null
+++ b/checkpoints/checkpoint-1000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:feeecf82b24e0ddf3c8d8285a678fe39a1184c1c961cf677b5ac8d36409a9a05
+size 1064
diff --git a/checkpoints/checkpoint-1000/trainer_state.json b/checkpoints/checkpoint-1000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..390fc3ea2539619e97f2debda2245ee0140a82df
--- /dev/null
+++ b/checkpoints/checkpoint-1000/trainer_state.json
@@ -0,0 +1,721 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1015950421619425,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-1000/training_args.bin b/checkpoints/checkpoint-1000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-1000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-1500/.DS_Store b/checkpoints/checkpoint-1500/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..5008ddfcf53c02e82d7eee2e57c38e5672ef89f6
Binary files /dev/null and b/checkpoints/checkpoint-1500/.DS_Store differ
diff --git a/checkpoints/checkpoint-1500/adapter_config.json b/checkpoints/checkpoint-1500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-1500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-1500/adapter_model.safetensors b/checkpoints/checkpoint-1500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..019ae5acc3c39c11dcf29cb8a850039b38bc7fa3
--- /dev/null
+++ b/checkpoints/checkpoint-1500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0429f41a7679282d393919c7b40c076ed2bde721f9f95c9c14f1f57fa0b63f6
+size 5919456
diff --git a/checkpoints/checkpoint-1500/optimizer.pt b/checkpoints/checkpoint-1500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b8ea20aac602d99eebb3facca793da7481fb829f
--- /dev/null
+++ b/checkpoints/checkpoint-1500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1cffc930b18a6507977b7928c70647d37429193d760799a24e4aa466c787fd3
+size 11930938
diff --git a/checkpoints/checkpoint-1500/rng_state_0.pth b/checkpoints/checkpoint-1500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..dd23fb6362f7699ea3bcc22f01e5b102e65d2118
--- /dev/null
+++ b/checkpoints/checkpoint-1500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af2583d0490e9c0c8a5832ae63a5d486d9951078b0ec9594ee0125a1807e7528
+size 15024
diff --git a/checkpoints/checkpoint-1500/rng_state_1.pth b/checkpoints/checkpoint-1500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d2851666bc50646272a2546dd1712a3eea0259bd
--- /dev/null
+++ b/checkpoints/checkpoint-1500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af9fd6e5a8f29754daaa4f1a3c57f904108d748a165ca0e1ad16571d90e39fa3
+size 15024
diff --git a/checkpoints/checkpoint-1500/rng_state_2.pth b/checkpoints/checkpoint-1500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..ac093b6104318e553a760be63eb06cd0d3f1ec17
--- /dev/null
+++ b/checkpoints/checkpoint-1500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd55da1784f3e2dba7244fb29d3bbc59fbefd6b4bb1357a4ded5822c60485304
+size 15024
diff --git a/checkpoints/checkpoint-1500/rng_state_3.pth b/checkpoints/checkpoint-1500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7ddbc69dcf0ed5d1b65deb88a8f1131eccc34e07
--- /dev/null
+++ b/checkpoints/checkpoint-1500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b10bab8fcd02c56978a924d6d0be691a36e321523009f8caa015318d52823f2
+size 15024
diff --git a/checkpoints/checkpoint-1500/scheduler.pt b/checkpoints/checkpoint-1500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..51876d3aaf0be1aae071cb0c017d51b61448d7ed
--- /dev/null
+++ b/checkpoints/checkpoint-1500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68a6cef91a2e578166a3e870fa8585312c20b665553a57976a77b1e7d2ca0ef3
+size 1064
diff --git a/checkpoints/checkpoint-1500/trainer_state.json b/checkpoints/checkpoint-1500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..228276cc158ac36aef26a0b87e0e57950412ba46
--- /dev/null
+++ b/checkpoints/checkpoint-1500/trainer_state.json
@@ -0,0 +1,1071 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.15239256324291375,
+  "eval_steps": 500,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-1500/training_args.bin b/checkpoints/checkpoint-1500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-1500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-2000/.DS_Store b/checkpoints/checkpoint-2000/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..5008ddfcf53c02e82d7eee2e57c38e5672ef89f6
Binary files /dev/null and b/checkpoints/checkpoint-2000/.DS_Store differ
diff --git a/checkpoints/checkpoint-2000/adapter_config.json b/checkpoints/checkpoint-2000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-2000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-2000/adapter_model.safetensors b/checkpoints/checkpoint-2000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4c42278d439b28a8eb84d8f3d7b93d55df223451
--- /dev/null
+++ b/checkpoints/checkpoint-2000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f511e364452e06b28c03299a544dc2d5fd730fdfd45007cd4ec530e30144139
+size 5919456
diff --git a/checkpoints/checkpoint-2000/optimizer.pt b/checkpoints/checkpoint-2000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..27718bc4fc2e46b271db6199e877b9ca8ddec162
--- /dev/null
+++ b/checkpoints/checkpoint-2000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4121e055d593f91209d1cfca077444bb962a8848054f778e04d22d7e269a91e3
+size 11930938
diff --git a/checkpoints/checkpoint-2000/rng_state_0.pth b/checkpoints/checkpoint-2000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3b1b550785550980db2dab7c4db776074d507397
--- /dev/null
+++ b/checkpoints/checkpoint-2000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7814979e8476866bd135a55e64d375b840c3c1436aa60c3d69ece3f0a10c3408
+size 15024
diff --git a/checkpoints/checkpoint-2000/rng_state_1.pth b/checkpoints/checkpoint-2000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7a3ed4fbc5afe1f89fd45284b352bf27deedb8ea
--- /dev/null
+++ b/checkpoints/checkpoint-2000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4b8a9f5b33e3bc1f4b31217176bbabc65ace6c56a7bf77b1b7153dc062ba709
+size 15024
diff --git a/checkpoints/checkpoint-2000/rng_state_2.pth b/checkpoints/checkpoint-2000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..6f9d319e7d0f09d65d040354c5e0320b59dec0f0
--- /dev/null
+++ b/checkpoints/checkpoint-2000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70445c98497e5f1d7bea8f44a93bc0211d0226177b834ac2949fe24e3b538d05
+size 15024
diff --git a/checkpoints/checkpoint-2000/rng_state_3.pth b/checkpoints/checkpoint-2000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..c23f68acc06264aeb9b09bf60269640ce8b147c8
--- /dev/null
+++ b/checkpoints/checkpoint-2000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c244dce3fafef8abbf503718fb81319ff12831928a8845134b78a845e0c6e14
+size 15024
diff --git a/checkpoints/checkpoint-2000/scheduler.pt b/checkpoints/checkpoint-2000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c133553a2b6f7143f327ed61a33a83fcc4a43b74
--- /dev/null
+++ b/checkpoints/checkpoint-2000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9cf822c73733a63e3bebbb2802abc0acedf3d9423d9f25c6bb103f490864c06
+size 1064
diff --git a/checkpoints/checkpoint-2000/trainer_state.json b/checkpoints/checkpoint-2000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e7a9301a9ac4e7f3efde74d978b93fbd74160b85
--- /dev/null
+++ b/checkpoints/checkpoint-2000/trainer_state.json
@@ -0,0 +1,1421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.203190084323885,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-2000/training_args.bin b/checkpoints/checkpoint-2000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-2000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-2500/.DS_Store b/checkpoints/checkpoint-2500/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..5008ddfcf53c02e82d7eee2e57c38e5672ef89f6
Binary files /dev/null and b/checkpoints/checkpoint-2500/.DS_Store differ
diff --git a/checkpoints/checkpoint-2500/adapter_config.json b/checkpoints/checkpoint-2500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-2500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-2500/adapter_model.safetensors b/checkpoints/checkpoint-2500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..b72c393034479ae51f68bfdb488baeb0eca04321
--- /dev/null
+++ b/checkpoints/checkpoint-2500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56179446117ddaff95696487189a032d48a215d8456339c2ae5eda1870df93b9
+size 5919456
diff --git a/checkpoints/checkpoint-2500/optimizer.pt b/checkpoints/checkpoint-2500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0b52ed1ee54b291770f15300bf140925bca02fde
--- /dev/null
+++ b/checkpoints/checkpoint-2500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd04baeeb8fe514953f04bca66abb944e11003693bffa63c977e38849273cb49
+size 11930938
diff --git a/checkpoints/checkpoint-2500/rng_state_0.pth b/checkpoints/checkpoint-2500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..17380c6e5ac5082c526e6f93eccf203432124116
--- /dev/null
+++ b/checkpoints/checkpoint-2500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:460a81dc2425be030afb2c9930b7e9fc9de54ad9cf988c330851fefe47a118c2
+size 15024
diff --git a/checkpoints/checkpoint-2500/rng_state_1.pth b/checkpoints/checkpoint-2500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..38dd76771028ab39d8a7742325c670728c9d9b17
--- /dev/null
+++ b/checkpoints/checkpoint-2500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f39c713bcb800d5d9d55e44b2d2e744bbee1e449a935b9a681868e507ac58f86
+size 15024
diff --git a/checkpoints/checkpoint-2500/rng_state_2.pth b/checkpoints/checkpoint-2500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..ba8fc1fc2b8078c75a9a126da3a3f68cd3411b42
--- /dev/null
+++ b/checkpoints/checkpoint-2500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6b1c81cf4c5daaf0fe1a5aa4e87259ede9f631f0867a6d085b8eba9a03f7275
+size 15024
diff --git a/checkpoints/checkpoint-2500/rng_state_3.pth b/checkpoints/checkpoint-2500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7f389e22cd4c2984db2f53ef472201c762aff9e6
--- /dev/null
+++ b/checkpoints/checkpoint-2500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7ae3d71695a016b734120d0e84d8510429872cccd4cdfc5ad051249bdceb709
+size 15024
diff --git a/checkpoints/checkpoint-2500/scheduler.pt b/checkpoints/checkpoint-2500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cfe96bf12bb338be15a0e572c935955e022dddb6
--- /dev/null
+++ b/checkpoints/checkpoint-2500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0992ea36b796a9efd875f53514a1a7b72426fb94846549bcad84ea3eae0acee
+size 1064
diff --git a/checkpoints/checkpoint-2500/trainer_state.json b/checkpoints/checkpoint-2500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..085f59bbcf897cba748b6ce5aee019fdbd65bb0d
--- /dev/null
+++ b/checkpoints/checkpoint-2500/trainer_state.json
@@ -0,0 +1,1771 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.25398760540485626,
+  "eval_steps": 500,
+  "global_step": 2500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-2500/training_args.bin b/checkpoints/checkpoint-2500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-2500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-3000/.DS_Store b/checkpoints/checkpoint-3000/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..5008ddfcf53c02e82d7eee2e57c38e5672ef89f6
Binary files /dev/null and b/checkpoints/checkpoint-3000/.DS_Store differ
diff --git a/checkpoints/checkpoint-3000/adapter_config.json b/checkpoints/checkpoint-3000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-3000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-3000/adapter_model.safetensors b/checkpoints/checkpoint-3000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f00b979e893c419aa35f668c6083983f38a702aa
--- /dev/null
+++ b/checkpoints/checkpoint-3000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24a237f0667aea50351b9a62bb2ddead2e09a567815c526d250be96397fa3798
+size 5919456
diff --git a/checkpoints/checkpoint-3000/optimizer.pt b/checkpoints/checkpoint-3000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5dfd65a95d4b75ec54a7c63d0fd72ce2529b8de1
--- /dev/null
+++ b/checkpoints/checkpoint-3000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb7f56c657475efd0fee1d492240e0c5e5d10ed8ec46630177e04192f2950d28
+size 11930938
diff --git a/checkpoints/checkpoint-3000/rng_state_0.pth b/checkpoints/checkpoint-3000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fd0da84997e5db079fe7df2a31aeef3efb336831
--- /dev/null
+++ b/checkpoints/checkpoint-3000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75565d94ecdf86007209f425495e6767052cfd0684203dcaf72c23b7c2dc2740
+size 15024
diff --git a/checkpoints/checkpoint-3000/rng_state_1.pth b/checkpoints/checkpoint-3000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..9582aaaef271b2855ca000b54510b991df4f866d
--- /dev/null
+++ b/checkpoints/checkpoint-3000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f986f3238448cdfd898312547f70e6317c7c4c40d4f49e7782f5380769c5f64
+size 15024
diff --git a/checkpoints/checkpoint-3000/rng_state_2.pth b/checkpoints/checkpoint-3000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0673c505e8967506093c68f5b6c6140a292635b5
--- /dev/null
+++ b/checkpoints/checkpoint-3000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:113f522722f00018489ce132b51c24506ef9b8b849c0c2a783913565b35f6cba
+size 15024
diff --git a/checkpoints/checkpoint-3000/rng_state_3.pth b/checkpoints/checkpoint-3000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..af1dca7b2f752be096e9e7b1b84f0d1897e9f780
--- /dev/null
+++ b/checkpoints/checkpoint-3000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1134f64cfe9f598e01f68bfb6bc3e69460c4b0c7ef7617bf88d7f539babada99
+size 15024
diff --git a/checkpoints/checkpoint-3000/scheduler.pt b/checkpoints/checkpoint-3000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..78eed6ed36accb49b60ba0c6db316b6bf427c96b
--- /dev/null
+++ b/checkpoints/checkpoint-3000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c5a5e7ee1063db317ee978d346cf3b726f5d9858cb8bd50568280ada9f34910
+size 1064
diff --git a/checkpoints/checkpoint-3000/trainer_state.json b/checkpoints/checkpoint-3000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..b813032acb2fc29774636a82567e34aeaaddb1b4
--- /dev/null
+++ b/checkpoints/checkpoint-3000/trainer_state.json
@@ -0,0 +1,2121 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.3047851264858275,
+  "eval_steps": 500,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-3000/training_args.bin b/checkpoints/checkpoint-3000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-3000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-3500/adapter_config.json b/checkpoints/checkpoint-3500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-3500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-3500/adapter_model.safetensors b/checkpoints/checkpoint-3500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..def26c035ec01dbe6399e1deda89633a3951ba73
--- /dev/null
+++ b/checkpoints/checkpoint-3500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbf46c695f24f5160623672ddb7b6e3230a79b1c62b8448e3dedb83005ca9935
+size 5919456
diff --git a/checkpoints/checkpoint-3500/optimizer.pt b/checkpoints/checkpoint-3500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5c107db7039ba7af8dcfa32fd7aa7a8fbd21c74f
--- /dev/null
+++ b/checkpoints/checkpoint-3500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22f791945b5138ee83b222ac582e4f208ff52e423ba51084873ae89d31ac1845
+size 11930938
diff --git a/checkpoints/checkpoint-3500/rng_state_0.pth b/checkpoints/checkpoint-3500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..6e2e5670f0b559873d46a22a975469d24350dae5
--- /dev/null
+++ b/checkpoints/checkpoint-3500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:937c0b4f9e0770f6eca6c4cdbab48674491808073f9a0008242c7747205b4b0a
+size 15024
diff --git a/checkpoints/checkpoint-3500/rng_state_1.pth b/checkpoints/checkpoint-3500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7eb9d5104e7e96e55a0cd774e3249194f48334fc
--- /dev/null
+++ b/checkpoints/checkpoint-3500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db86a3a2768366792f057841f721820dff6815adc16d464af7828df941223c01
+size 15024
diff --git a/checkpoints/checkpoint-3500/rng_state_2.pth b/checkpoints/checkpoint-3500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fda380880dfe7ab57f96a6a7914e3ae9c2b57485
--- /dev/null
+++ b/checkpoints/checkpoint-3500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fc2e3f389bdedf3f7c940ce7cfd74a54d5ef411c9271a0b570f0f8078360d61
+size 15024
diff --git a/checkpoints/checkpoint-3500/rng_state_3.pth b/checkpoints/checkpoint-3500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..8422b115f583e37a495ec690fab1f7a2441ec0a8
--- /dev/null
+++ b/checkpoints/checkpoint-3500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:700a170dc517eaa370c876455bedd04e0b25e7140ee505e562880ac0c268e199
+size 15024
diff --git a/checkpoints/checkpoint-3500/scheduler.pt b/checkpoints/checkpoint-3500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4c79541e6a9cb3467a5d3ceaa6d2065ec4525bfb
--- /dev/null
+++ b/checkpoints/checkpoint-3500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:586f3b2ba8da5de2fa8f36edc9313f3dbd8056e41828a6210b6c8e0807689cc2
+size 1064
diff --git a/checkpoints/checkpoint-3500/trainer_state.json b/checkpoints/checkpoint-3500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..d8f21bb043f14ed3122d67861c839124c000b40f
--- /dev/null
+++ b/checkpoints/checkpoint-3500/trainer_state.json
@@ -0,0 +1,2471 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.35558264756679875,
+  "eval_steps": 500,
+  "global_step": 3500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-3500/training_args.bin b/checkpoints/checkpoint-3500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-3500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-4000/adapter_config.json b/checkpoints/checkpoint-4000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-4000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-4000/adapter_model.safetensors b/checkpoints/checkpoint-4000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0010e62e10adb58b59a4171d87f3fa37c74f4b27
--- /dev/null
+++ b/checkpoints/checkpoint-4000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bfc0a45e20d761b2e036c7f419351073780fc963f478f6726d61bf6913308b5
+size 5919456
diff --git a/checkpoints/checkpoint-4000/optimizer.pt b/checkpoints/checkpoint-4000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7c063fe1aecc2f351c80b43394373815b139c4c7
--- /dev/null
+++ b/checkpoints/checkpoint-4000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:030a97b642b9ca1bca5998b5c77cfdcf063fd297b4aded2538f62462ffcb0d5f
+size 11930938
diff --git a/checkpoints/checkpoint-4000/rng_state_0.pth b/checkpoints/checkpoint-4000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5c7dcab2c1fa96e39da4e4616e5e8b282a2f6923
--- /dev/null
+++ b/checkpoints/checkpoint-4000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdcaaa7cfa8093e078769990ecfb571790fea6d1f1143531eadfe0ef4d53d941
+size 15024
diff --git a/checkpoints/checkpoint-4000/rng_state_1.pth b/checkpoints/checkpoint-4000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..db8664d3862d07cb99bf09b73c3eb39bf3a8cfb0
--- /dev/null
+++ b/checkpoints/checkpoint-4000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1e818c945017fe6ffde32b0e04843796059b044ff660b74cf4c5e8397ef5bc3
+size 15024
diff --git a/checkpoints/checkpoint-4000/rng_state_2.pth b/checkpoints/checkpoint-4000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..485a59f7e6666a852992eb89a47736cd878a4cf8
--- /dev/null
+++ b/checkpoints/checkpoint-4000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2847ec9ff5d020b7601c16b4794123d4fc5216fa782ba5350f8d8d42d63dec99
+size 15024
diff --git a/checkpoints/checkpoint-4000/rng_state_3.pth b/checkpoints/checkpoint-4000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..a9f9eb06dbb4d30c19a4bbda9fb0234299c087a7
--- /dev/null
+++ b/checkpoints/checkpoint-4000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d3b7dd272ec99203448b0a7997c01b2d5e414ac34fcaf4a77f23495f1864166
+size 15024
diff --git a/checkpoints/checkpoint-4000/scheduler.pt b/checkpoints/checkpoint-4000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2f9cc3da0d207e2f0ffb59d7ccbff46198b1d23f
--- /dev/null
+++ b/checkpoints/checkpoint-4000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa29240df220021e7e33aee35824e879cf6ebdc5ef315b12a0da849e15ca9816
+size 1064
diff --git a/checkpoints/checkpoint-4000/trainer_state.json b/checkpoints/checkpoint-4000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f1ed37679ac992e6c6dcdfdce95ee04973dd71d7
--- /dev/null
+++ b/checkpoints/checkpoint-4000/trainer_state.json
@@ -0,0 +1,2821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.40638016864777,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-4000/training_args.bin b/checkpoints/checkpoint-4000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-4000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-4500/adapter_config.json b/checkpoints/checkpoint-4500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-4500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-4500/adapter_model.safetensors b/checkpoints/checkpoint-4500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7aa3071a8b60a1911581ad4fc94db44a264cd1e1
--- /dev/null
+++ b/checkpoints/checkpoint-4500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7c6963c0425b0c2a06a71e6d47d4fbd98e2a133e8d06948e611eee31aa731b5
+size 5919456
diff --git a/checkpoints/checkpoint-4500/optimizer.pt b/checkpoints/checkpoint-4500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d0cf2fc1750b9ba6635b3270b2477e9c69023282
--- /dev/null
+++ b/checkpoints/checkpoint-4500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0ca7ae9cafbccffe17ee656b749d23931f579d38d02f6c6b7b74c6a48567952
+size 11930938
diff --git a/checkpoints/checkpoint-4500/rng_state_0.pth b/checkpoints/checkpoint-4500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..79c43bfab83f7a282b4b114757cd92312afef77e
--- /dev/null
+++ b/checkpoints/checkpoint-4500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7db56d699f6c4a471d74081289b1dc738167e2c8cb96555bccff59ced1cca60e
+size 15024
diff --git a/checkpoints/checkpoint-4500/rng_state_1.pth b/checkpoints/checkpoint-4500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0c66d8f438b500c92afd06878b0368f9291d1969
--- /dev/null
+++ b/checkpoints/checkpoint-4500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcf572c6921c759e3118791b83271fd15d5f57461b5f6071b7e35d40b08d7c33
+size 15024
diff --git a/checkpoints/checkpoint-4500/rng_state_2.pth b/checkpoints/checkpoint-4500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b19443304b49a814d70104106104e9252ab03822
--- /dev/null
+++ b/checkpoints/checkpoint-4500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce499e2b0e0eea0c44c5e4fe25e271573ced5a8fefd0acff896293acd6102969
+size 15024
diff --git a/checkpoints/checkpoint-4500/rng_state_3.pth b/checkpoints/checkpoint-4500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..118e3785c401a599c8cce9ab1d24766b140cee99
--- /dev/null
+++ b/checkpoints/checkpoint-4500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b4a17f4cb7c3a2fc3e2abbc1b812e3de44942fd00e096dc1860327c3e174cf1
+size 15024
diff --git a/checkpoints/checkpoint-4500/scheduler.pt b/checkpoints/checkpoint-4500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3cf7e49f7f8f9c6c02c1fea57551f9b19ff36b35
--- /dev/null
+++ b/checkpoints/checkpoint-4500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:157e40aa454761c5046fdbfd6ead1c155f32a8d8a91eba044250c5c2d1b7f7fa
+size 1064
diff --git a/checkpoints/checkpoint-4500/trainer_state.json b/checkpoints/checkpoint-4500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f17c9ab3c88953159e52b1f4f07808a0b09ac082
--- /dev/null
+++ b/checkpoints/checkpoint-4500/trainer_state.json
@@ -0,0 +1,3171 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.45717768972874123,
+  "eval_steps": 500,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-4500/training_args.bin b/checkpoints/checkpoint-4500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-4500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-500/.DS_Store b/checkpoints/checkpoint-500/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..5008ddfcf53c02e82d7eee2e57c38e5672ef89f6
Binary files /dev/null and b/checkpoints/checkpoint-500/.DS_Store differ
diff --git a/checkpoints/checkpoint-500/adapter_config.json b/checkpoints/checkpoint-500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-500/adapter_model.safetensors b/checkpoints/checkpoint-500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e9303fa912a4ef3e12e8166a7375958634ce8f36
--- /dev/null
+++ b/checkpoints/checkpoint-500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe875d0568aa37f2927a8a3eb8f6951cb2a06db4bb4acf3f08994191b8fcd074
+size 5919456
diff --git a/checkpoints/checkpoint-500/optimizer.pt b/checkpoints/checkpoint-500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..df3daed252c5700e51cc06cd0106fb5befa94f45
--- /dev/null
+++ b/checkpoints/checkpoint-500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcefad058bb98bc378f909e2924345eaa383cbd31ac20a7c96a0144fbdad481b
+size 11930938
diff --git a/checkpoints/checkpoint-500/rng_state_0.pth b/checkpoints/checkpoint-500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5e831911e803d34db0dee43e81dbe00dd8ce41a1
--- /dev/null
+++ b/checkpoints/checkpoint-500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:809bd09d083f3cc7af22a0d5fa482e3b4a80ee095c2d25606491adff7d437298
+size 15024
diff --git a/checkpoints/checkpoint-500/rng_state_1.pth b/checkpoints/checkpoint-500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5fc931d40d8b42b24ee9524277102c98e47a7f1c
--- /dev/null
+++ b/checkpoints/checkpoint-500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f59c5bba33ae6a197dd330c2dad3e367df715420869ffee03f041f6bf374bd47
+size 15024
diff --git a/checkpoints/checkpoint-500/rng_state_2.pth b/checkpoints/checkpoint-500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..db1a2781c331b27ce633c771e2eb5ef32c951b3b
--- /dev/null
+++ b/checkpoints/checkpoint-500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55e1bb7909b9abc9e4e3f3acf48720cbfc308fbfb27de6a9043a72cfc99abc32
+size 15024
diff --git a/checkpoints/checkpoint-500/rng_state_3.pth b/checkpoints/checkpoint-500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..58211359f16a7adad0b07c2acd96237202353fe7
--- /dev/null
+++ b/checkpoints/checkpoint-500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ade95ac7fe747f684693248dcde4845f243707debd6487625345278fdd7ad55
+size 15024
diff --git a/checkpoints/checkpoint-500/scheduler.pt b/checkpoints/checkpoint-500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f50d6e23087c0212f07d2b373bd3914a857dbcf4
--- /dev/null
+++ b/checkpoints/checkpoint-500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9494fbf9d2b6d20c7d21fb5ce991f87947f9899a55c5171fe969b0e4974d2103
+size 1064
diff --git a/checkpoints/checkpoint-500/trainer_state.json b/checkpoints/checkpoint-500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..b0d2b5be8b9988aeafaf4997b3b16453e990e766
--- /dev/null
+++ b/checkpoints/checkpoint-500/trainer_state.json
@@ -0,0 +1,371 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.05079752108097125,
+  "eval_steps": 500,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-500/training_args.bin b/checkpoints/checkpoint-500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-5000/adapter_config.json b/checkpoints/checkpoint-5000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-5000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-5000/adapter_model.safetensors b/checkpoints/checkpoint-5000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c5323954094d8a6496c38b30ffa0df0ac0cfffe8
--- /dev/null
+++ b/checkpoints/checkpoint-5000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ae5f313d7b029213719465291718ddf1e014b263490e78fc2a5169bf0b5252a
+size 5919456
diff --git a/checkpoints/checkpoint-5000/optimizer.pt b/checkpoints/checkpoint-5000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a79946ce5dd1275af1622625dd5f1f7e9f277b2d
--- /dev/null
+++ b/checkpoints/checkpoint-5000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12db3a411b1a777bb2de716e77eb87fa0eb100990039bc8de878ad51ab65b732
+size 11930938
diff --git a/checkpoints/checkpoint-5000/rng_state_0.pth b/checkpoints/checkpoint-5000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3af645b6e41cbac80e3d515e0208c0dfd154af79
--- /dev/null
+++ b/checkpoints/checkpoint-5000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67bfaa3beff54b75efd8c9a44ed5971676113e2db1f831aa1c6af1ce6a9caa0b
+size 15024
diff --git a/checkpoints/checkpoint-5000/rng_state_1.pth b/checkpoints/checkpoint-5000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..597c43e1460e8c299d649484b08ee0173163f16f
--- /dev/null
+++ b/checkpoints/checkpoint-5000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41ed0b09eacb9d8a8c6e72cf8f95eee4f179b7f40a1ead1975ec33f39a10b112
+size 15024
diff --git a/checkpoints/checkpoint-5000/rng_state_2.pth b/checkpoints/checkpoint-5000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..c826825016b292604ccf1c156229766246fb116e
--- /dev/null
+++ b/checkpoints/checkpoint-5000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8d837f21251c93900eb66777d51bff1fbc27c649b7852fa3ff8933fb4214304
+size 15024
diff --git a/checkpoints/checkpoint-5000/rng_state_3.pth b/checkpoints/checkpoint-5000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..c344167e5746676b6dde2a30375e5d12f6600233
--- /dev/null
+++ b/checkpoints/checkpoint-5000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efef25c82d2ac74d69a7a158e6fb9e4ce251ed5ebd2235b890bf805ca5dc093e
+size 15024
diff --git a/checkpoints/checkpoint-5000/scheduler.pt b/checkpoints/checkpoint-5000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c0604ddc2c73d6b2fa706115b186b8876b4db0f1
--- /dev/null
+++ b/checkpoints/checkpoint-5000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee42deb5c1610056c95ecf4b744f4223a6749192630e5e7100203f9bf0540ef4
+size 1064
diff --git a/checkpoints/checkpoint-5000/trainer_state.json b/checkpoints/checkpoint-5000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..360c1e39d3d8b26d00d06e5b1ec148375669ff0d
--- /dev/null
+++ b/checkpoints/checkpoint-5000/trainer_state.json
@@ -0,0 +1,3521 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5079752108097125,
+  "eval_steps": 500,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-5000/training_args.bin b/checkpoints/checkpoint-5000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-5000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-5500/adapter_config.json b/checkpoints/checkpoint-5500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-5500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-5500/adapter_model.safetensors b/checkpoints/checkpoint-5500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..21091e5fb953f81db93b5a0a1b0e35419d285ff5
--- /dev/null
+++ b/checkpoints/checkpoint-5500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb810b92dc1e55cf94b723b80404b69163c821876b612daca7c18abd50e5cc29
+size 5919456
diff --git a/checkpoints/checkpoint-5500/optimizer.pt b/checkpoints/checkpoint-5500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..283b36631e755fd00b60aa6cd5641a403ac76671
--- /dev/null
+++ b/checkpoints/checkpoint-5500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf105da9fd77aa95ac72217635110bc055276ac42b13c61c0ea8afe029df437c
+size 11930938
diff --git a/checkpoints/checkpoint-5500/rng_state_0.pth b/checkpoints/checkpoint-5500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7b48e0fe6cf94e78b55f9c79e0ea7aa19d768484
--- /dev/null
+++ b/checkpoints/checkpoint-5500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70343f9cb54717109b4b086674f4317c77a6d5236e01a3f60ae87d47f6637943
+size 15024
diff --git a/checkpoints/checkpoint-5500/rng_state_1.pth b/checkpoints/checkpoint-5500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..f870e3cd3d0baad3b6137f60d4d1eed197e6b688
--- /dev/null
+++ b/checkpoints/checkpoint-5500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17e6a7bceffa2b144c23b4b2c0efacf64ed5a2f0808ff897d8e65e0e129c709d
+size 15024
diff --git a/checkpoints/checkpoint-5500/rng_state_2.pth b/checkpoints/checkpoint-5500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..a3f7b4465900221f00761bcc0d874e660643e438
--- /dev/null
+++ b/checkpoints/checkpoint-5500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:109af0ffa2fb7f677448e11bba29f896c314794923e50f5ab77002d63db44682
+size 15024
diff --git a/checkpoints/checkpoint-5500/rng_state_3.pth b/checkpoints/checkpoint-5500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3912593f43081836761363a4b700970c467e59f3
--- /dev/null
+++ b/checkpoints/checkpoint-5500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:043288038a590b0ede4ede98760e44fe949c1db3d9630240030a93cb18c91259
+size 15024
diff --git a/checkpoints/checkpoint-5500/scheduler.pt b/checkpoints/checkpoint-5500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8b047cded6249de59455f00473cd79106488c266
--- /dev/null
+++ b/checkpoints/checkpoint-5500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5608c457217d0ace38955805f4007ad8ac45872c2d7cde753cbd1a93ae8304bd
+size 1064
diff --git a/checkpoints/checkpoint-5500/trainer_state.json b/checkpoints/checkpoint-5500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..d651783d2a32611ed806f44e48a3ee9af386d8df
--- /dev/null
+++ b/checkpoints/checkpoint-5500/trainer_state.json
@@ -0,0 +1,3871 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5587727318906838,
+  "eval_steps": 500,
+  "global_step": 5500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-5500/training_args.bin b/checkpoints/checkpoint-5500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-5500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-6000/adapter_config.json b/checkpoints/checkpoint-6000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-6000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-6000/adapter_model.safetensors b/checkpoints/checkpoint-6000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4952b4876cdc8d06ed1f82e75e7acfb784d7c302
--- /dev/null
+++ b/checkpoints/checkpoint-6000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69f4641ab691c0b74c8e8075b2ec23906859a6fe6d1279160c1d6f7281d2611e
+size 5919456
diff --git a/checkpoints/checkpoint-6000/optimizer.pt b/checkpoints/checkpoint-6000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8a6d428db6bdf0026167613f3f78e4cbb391cf2e
--- /dev/null
+++ b/checkpoints/checkpoint-6000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ccafd81d481629f8476e5ad9a763e445cfca7e7ec2aba5cffe2ed9ccacdb684
+size 11930938
diff --git a/checkpoints/checkpoint-6000/rng_state_0.pth b/checkpoints/checkpoint-6000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..856f6e848cfe511c4d3682477fc96768fa71e17b
--- /dev/null
+++ b/checkpoints/checkpoint-6000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf4e27664b653acfc21de70ae172ec0726ec640b898e117c9e038d403049764b
+size 15024
diff --git a/checkpoints/checkpoint-6000/rng_state_1.pth b/checkpoints/checkpoint-6000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..91439d971c1906d490e43d5ab3edec52712a346c
--- /dev/null
+++ b/checkpoints/checkpoint-6000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee19749a64079934bb3c47b0f06798649c23ef5b2d28a94a6161759a9b11f5ae
+size 15024
diff --git a/checkpoints/checkpoint-6000/rng_state_2.pth b/checkpoints/checkpoint-6000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..6426756daceb7e8531a5a3124f838933e0855709
--- /dev/null
+++ b/checkpoints/checkpoint-6000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1eb0b946551fdb70140918364d81243b8abd5ef8ab1df8ba0a040cd91d240e5
+size 15024
diff --git a/checkpoints/checkpoint-6000/rng_state_3.pth b/checkpoints/checkpoint-6000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..8c11f16945bb9db73e4cb93bc2cc5c8ebd465457
--- /dev/null
+++ b/checkpoints/checkpoint-6000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5684284bf0bd122be00905a7d1688bf38e87d37239ed219e780f5b8d8f5f3eb
+size 15024
diff --git a/checkpoints/checkpoint-6000/scheduler.pt b/checkpoints/checkpoint-6000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6e228e34327b45950a62a960d14420ebe108d897
--- /dev/null
+++ b/checkpoints/checkpoint-6000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:202173950488d1b3f4ccd086089dbc7f02c295eb11ea39c10e13fac52ece8f3a
+size 1064
diff --git a/checkpoints/checkpoint-6000/trainer_state.json b/checkpoints/checkpoint-6000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..8bfa9cc79f70409d1c63f3a37aa77adcf038e581
--- /dev/null
+++ b/checkpoints/checkpoint-6000/trainer_state.json
@@ -0,0 +1,4221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.609570252971655,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-6000/training_args.bin b/checkpoints/checkpoint-6000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-6000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-6500/adapter_config.json b/checkpoints/checkpoint-6500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-6500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-6500/adapter_model.safetensors b/checkpoints/checkpoint-6500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ea7be9ee67cb55626a3d8d3ef209fc4f1bbf148f
--- /dev/null
+++ b/checkpoints/checkpoint-6500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14ab2e0853dd44a4b03a9d15251deba57b15ed3261fc5e75ea0c78d8f9481486
+size 5919456
diff --git a/checkpoints/checkpoint-6500/optimizer.pt b/checkpoints/checkpoint-6500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5ce7862baad54ec5623f4b4c5b4996f11f14e8c1
--- /dev/null
+++ b/checkpoints/checkpoint-6500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7047b3d5af563c786060cdd116349c114053ee2307bebbbf5f5214200bcde1ce
+size 11930938
diff --git a/checkpoints/checkpoint-6500/rng_state_0.pth b/checkpoints/checkpoint-6500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0e8e09c2f4a00e6486fc03f4d8a2e8a9eb1caac7
--- /dev/null
+++ b/checkpoints/checkpoint-6500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a591de0feafd38f3dfb34a5639b5f07fc69364d8918cb41bee6dd6766d4ef1d2
+size 15024
diff --git a/checkpoints/checkpoint-6500/rng_state_1.pth b/checkpoints/checkpoint-6500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0b9a1717d7d349f476f13e5192f06dc234ca82ea
--- /dev/null
+++ b/checkpoints/checkpoint-6500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da06f3a5ac82888a33cdffca90c924defae45ba6fd9ff5004219e2c9f4170e79
+size 15024
diff --git a/checkpoints/checkpoint-6500/rng_state_2.pth b/checkpoints/checkpoint-6500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..504d07665b63dfacbdb864145313ac5c3fee5a57
--- /dev/null
+++ b/checkpoints/checkpoint-6500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d391446f8c14431e9c25329555ff1b3c6971fcfbfabf701b23c665e748c381f
+size 15024
diff --git a/checkpoints/checkpoint-6500/rng_state_3.pth b/checkpoints/checkpoint-6500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fcf06f5ee914106315184abc1f4e43a3e0dc6a3e
--- /dev/null
+++ b/checkpoints/checkpoint-6500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34b3a168af7b46a8a4885487b4c51f6f868f8ac39e0a20191646e4671ae739ea
+size 15024
diff --git a/checkpoints/checkpoint-6500/scheduler.pt b/checkpoints/checkpoint-6500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bed16b1cc961349a026d0a7300f4c19a4a4cb3b3
--- /dev/null
+++ b/checkpoints/checkpoint-6500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bdea953796bedc1929c6f94c7498a5dbf5c6f421163832ce66f06c7e038f707
+size 1064
diff --git a/checkpoints/checkpoint-6500/trainer_state.json b/checkpoints/checkpoint-6500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..1e3fe36bbb17bb1b607af62a6682ab9d45672e47
--- /dev/null
+++ b/checkpoints/checkpoint-6500/trainer_state.json
@@ -0,0 +1,4571 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6603677740526263,
+  "eval_steps": 500,
+  "global_step": 6500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-6500/training_args.bin b/checkpoints/checkpoint-6500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-6500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-7000/adapter_config.json b/checkpoints/checkpoint-7000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-7000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-7000/adapter_model.safetensors b/checkpoints/checkpoint-7000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f035c3b02bb414b55a51e0686d93c6e41268b717
--- /dev/null
+++ b/checkpoints/checkpoint-7000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcfbb907ba164ea8f4886b499272e552c75b66b5f57c2758a433e516afc4d4ce
+size 5919456
diff --git a/checkpoints/checkpoint-7000/optimizer.pt b/checkpoints/checkpoint-7000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e2b419628c564de6327192effa738d6dcc841ee6
--- /dev/null
+++ b/checkpoints/checkpoint-7000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18252d03a633e94bef652fd14050f4f4c005fa3e1408f904a8582bde8cf6615c
+size 11930938
diff --git a/checkpoints/checkpoint-7000/rng_state_0.pth b/checkpoints/checkpoint-7000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..8a64a4f50d135d9aab30a3efe7dd0f870f273099
--- /dev/null
+++ b/checkpoints/checkpoint-7000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:189e349a3350f8ede7259e8b9c3805a87a95924562b5e53a021c94a808a1c148
+size 15024
diff --git a/checkpoints/checkpoint-7000/rng_state_1.pth b/checkpoints/checkpoint-7000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3669897a1580ca7339f57897968f1cd06fa56704
--- /dev/null
+++ b/checkpoints/checkpoint-7000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb75d1d5d631f8db237b129d4d42263dc618eca39919111a64bdee685ea49d66
+size 15024
diff --git a/checkpoints/checkpoint-7000/rng_state_2.pth b/checkpoints/checkpoint-7000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..26649e78cb314240d04745c505919161570ff823
--- /dev/null
+++ b/checkpoints/checkpoint-7000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35edf2986023cc71305d54b71c695198c4e470f12b08318d2bdd928b6f0040bb
+size 15024
diff --git a/checkpoints/checkpoint-7000/rng_state_3.pth b/checkpoints/checkpoint-7000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..18f1294c99a3714188a7fd11fee212fba2a544af
--- /dev/null
+++ b/checkpoints/checkpoint-7000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6aeb7ee84fba724996f3c432c15837e87dd1f98cb887991d0b5902eadb092ea
+size 15024
diff --git a/checkpoints/checkpoint-7000/scheduler.pt b/checkpoints/checkpoint-7000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..38eb3798e21f4a386d944a1c7b7f72bbd3680866
--- /dev/null
+++ b/checkpoints/checkpoint-7000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c38f0818bc728352543f333573634bed4254978229e01cce08ff84863c12dc13
+size 1064
diff --git a/checkpoints/checkpoint-7000/trainer_state.json b/checkpoints/checkpoint-7000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..2be0b7f759106ae9ab905f59d43ffd30cde0054b
--- /dev/null
+++ b/checkpoints/checkpoint-7000/trainer_state.json
@@ -0,0 +1,4921 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7111652951335975,
+  "eval_steps": 500,
+  "global_step": 7000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6613837244742456,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.3100382346268392e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 0.662399674895865,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.3029550230900812e-05,
+      "loss": 0.1997,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6634156253174845,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2958842573996016e-05,
+      "loss": 0.1969,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6644315757391039,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2888260110711525e-05,
+      "loss": 0.1469,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6654475261607233,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.2817803574903212e-05,
+      "loss": 0.1524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6664634765823427,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2747473699117668e-05,
+      "loss": 0.159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6674794270039622,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.267727121458458e-05,
+      "loss": 0.1999,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6684953774255816,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.2607196851209137e-05,
+      "loss": 0.2216,
+      "step": 6580
+    },
+    {
+      "epoch": 0.669511327847201,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2537251337564412e-05,
+      "loss": 0.1607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6705272782688205,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.2467435400883839e-05,
+      "loss": 0.2187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6715432286904399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.239774976705359e-05,
+      "loss": 0.1753,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6725591791120593,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2328195160605092e-05,
+      "loss": 0.194,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6735751295336787,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.225877230470743e-05,
+      "loss": 0.1485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6745910799552982,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.218948192115988e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6756070303769176,
+      "grad_norm": 3.875,
+      "learning_rate": 1.21203247303844e-05,
+      "loss": 0.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.676622980798537,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.2051301451418073e-05,
+      "loss": 0.2377,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6776389312201565,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.198241280190574e-05,
+      "loss": 0.1508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6786548816417759,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.1913659498092431e-05,
+      "loss": 0.1537,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6796708320633953,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.184504225481601e-05,
+      "loss": 0.2339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6806867824850147,
+      "grad_norm": 5.625,
+      "learning_rate": 1.177656178549966e-05,
+      "loss": 0.2102,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6817027329066342,
+      "grad_norm": 2.5,
+      "learning_rate": 1.1708218802144536e-05,
+      "loss": 0.1435,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6827186833282536,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1640014015322323e-05,
+      "loss": 0.1823,
+      "step": 6720
+    },
+    {
+      "epoch": 0.683734633749873,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1571948134167862e-05,
+      "loss": 0.1154,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6847505841714925,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1504021866371761e-05,
+      "loss": 0.2105,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6857665345931119,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.143623591817304e-05,
+      "loss": 0.1317,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6867824850147313,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.1368590994351835e-05,
+      "loss": 0.1406,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6877984354363507,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.130108779822198e-05,
+      "loss": 0.1425,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6888143858579702,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.1233727031623783e-05,
+      "loss": 0.1623,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6898303362795896,
+      "grad_norm": 4.625,
+      "learning_rate": 1.1166509394916682e-05,
+      "loss": 0.1591,
+      "step": 6790
+    },
+    {
+      "epoch": 0.690846286701209,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1099435586971982e-05,
+      "loss": 0.1758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6918622371228285,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.1032506305165555e-05,
+      "loss": 0.1018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6928781875444479,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.0965722245370641e-05,
+      "loss": 0.1485,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6938941379660672,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0899084101950561e-05,
+      "loss": 0.1762,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6949100883876866,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.0832592567751555e-05,
+      "loss": 0.1402,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6959260388093061,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0766248334095505e-05,
+      "loss": 0.2278,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6969419892309255,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.0700052090772828e-05,
+      "loss": 0.1969,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6979579396525449,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.0634004526035249e-05,
+      "loss": 0.2073,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6989738900741643,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.0568106326588645e-05,
+      "loss": 0.1902,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6999898404957838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.0502358177585953e-05,
+      "loss": 0.2165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7010057909174032,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.0436760762619977e-05,
+      "loss": 0.1952,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7020217413390226,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.0371314763716347e-05,
+      "loss": 0.1422,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7030376917606421,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.0306020861326388e-05,
+      "loss": 0.0961,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7040536421822615,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.0240879734320068e-05,
+      "loss": 0.1542,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7050695926038809,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.0175892059978901e-05,
+      "loss": 0.1748,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7060855430255003,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.0111058513988958e-05,
+      "loss": 0.0819,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7071014934471198,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.0046379770433803e-05,
+      "loss": 0.1933,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7081174438687392,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.98185650178749e-06,
+      "loss": 0.1891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7091333942903586,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.917489378907591e-06,
+      "loss": 0.2102,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7101493447119781,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.853279071028212e-06,
+      "loss": 0.1714,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7111652951335975,
+      "grad_norm": 2.375,
+      "learning_rate": 9.78922624575303e-06,
+      "loss": 0.1299,
+      "step": 7000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-7000/training_args.bin b/checkpoints/checkpoint-7000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-7000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-7500/adapter_config.json b/checkpoints/checkpoint-7500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-7500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-7500/adapter_model.safetensors b/checkpoints/checkpoint-7500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8b566b27331cac4819b9e695fbf784bff8e79cd5
--- /dev/null
+++ b/checkpoints/checkpoint-7500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eea7fbe808952b90c42a26bb6a156e294c6b01b74f4561e32845a867894829fc
+size 5919456
diff --git a/checkpoints/checkpoint-7500/optimizer.pt b/checkpoints/checkpoint-7500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fd01feb285dc960cdf690295976b4050319066c1
--- /dev/null
+++ b/checkpoints/checkpoint-7500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5dd7bb44df6f3a93ddba0c17a7309fbe69b5d9611641b66779f565c2707e80dd
+size 11930938
diff --git a/checkpoints/checkpoint-7500/rng_state_0.pth b/checkpoints/checkpoint-7500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0b99b3affc0b088b35c4713d5dbf363b6fb09e01
--- /dev/null
+++ b/checkpoints/checkpoint-7500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81a176e4f417232286c66488b5554a046fc1af84e1b5eff446ad37a4dc31c907
+size 15024
diff --git a/checkpoints/checkpoint-7500/rng_state_1.pth b/checkpoints/checkpoint-7500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..9e56badfa0d1b294971878e842fe14a8595016f1
--- /dev/null
+++ b/checkpoints/checkpoint-7500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3899c82e743501e2db6cd0e409a1b646a2eae511ad1e21aa77bc3f897edbf07a
+size 15024
diff --git a/checkpoints/checkpoint-7500/rng_state_2.pth b/checkpoints/checkpoint-7500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d5c75385d7ded955107873c0bccd2a63bd21f7ce
--- /dev/null
+++ b/checkpoints/checkpoint-7500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e12f41676d53322341924bbe6951161bec80e7d0e4adb780e12732cfa714e98d
+size 15024
diff --git a/checkpoints/checkpoint-7500/rng_state_3.pth b/checkpoints/checkpoint-7500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0f0ce3c448fcfd7f4444003295f128c63d9e46cd
--- /dev/null
+++ b/checkpoints/checkpoint-7500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b87c6fdf947fac25cf632f98a25e1c5f87369325d38ab0c3dfe29dea4f62eb75
+size 15024
diff --git a/checkpoints/checkpoint-7500/scheduler.pt b/checkpoints/checkpoint-7500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5a8d5e25619fe365bd3104d2c93dae8be1ab1f87
--- /dev/null
+++ b/checkpoints/checkpoint-7500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01aa4970faf173b76d7b990defe083ac27fc2fbe404a2ecca54f2098c3a5e177
+size 1064
diff --git a/checkpoints/checkpoint-7500/trainer_state.json b/checkpoints/checkpoint-7500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ebe79f3a10531feb2f6b9e25d7e9001aa29a39fc
--- /dev/null
+++ b/checkpoints/checkpoint-7500/trainer_state.json
@@ -0,0 +1,5271 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7619628162145687,
+  "eval_steps": 500,
+  "global_step": 7500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6613837244742456,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.3100382346268392e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 0.662399674895865,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.3029550230900812e-05,
+      "loss": 0.1997,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6634156253174845,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2958842573996016e-05,
+      "loss": 0.1969,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6644315757391039,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2888260110711525e-05,
+      "loss": 0.1469,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6654475261607233,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.2817803574903212e-05,
+      "loss": 0.1524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6664634765823427,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2747473699117668e-05,
+      "loss": 0.159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6674794270039622,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.267727121458458e-05,
+      "loss": 0.1999,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6684953774255816,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.2607196851209137e-05,
+      "loss": 0.2216,
+      "step": 6580
+    },
+    {
+      "epoch": 0.669511327847201,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2537251337564412e-05,
+      "loss": 0.1607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6705272782688205,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.2467435400883839e-05,
+      "loss": 0.2187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6715432286904399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.239774976705359e-05,
+      "loss": 0.1753,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6725591791120593,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2328195160605092e-05,
+      "loss": 0.194,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6735751295336787,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.225877230470743e-05,
+      "loss": 0.1485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6745910799552982,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.218948192115988e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6756070303769176,
+      "grad_norm": 3.875,
+      "learning_rate": 1.21203247303844e-05,
+      "loss": 0.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.676622980798537,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.2051301451418073e-05,
+      "loss": 0.2377,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6776389312201565,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.198241280190574e-05,
+      "loss": 0.1508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6786548816417759,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.1913659498092431e-05,
+      "loss": 0.1537,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6796708320633953,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.184504225481601e-05,
+      "loss": 0.2339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6806867824850147,
+      "grad_norm": 5.625,
+      "learning_rate": 1.177656178549966e-05,
+      "loss": 0.2102,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6817027329066342,
+      "grad_norm": 2.5,
+      "learning_rate": 1.1708218802144536e-05,
+      "loss": 0.1435,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6827186833282536,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1640014015322323e-05,
+      "loss": 0.1823,
+      "step": 6720
+    },
+    {
+      "epoch": 0.683734633749873,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1571948134167862e-05,
+      "loss": 0.1154,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6847505841714925,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1504021866371761e-05,
+      "loss": 0.2105,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6857665345931119,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.143623591817304e-05,
+      "loss": 0.1317,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6867824850147313,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.1368590994351835e-05,
+      "loss": 0.1406,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6877984354363507,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.130108779822198e-05,
+      "loss": 0.1425,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6888143858579702,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.1233727031623783e-05,
+      "loss": 0.1623,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6898303362795896,
+      "grad_norm": 4.625,
+      "learning_rate": 1.1166509394916682e-05,
+      "loss": 0.1591,
+      "step": 6790
+    },
+    {
+      "epoch": 0.690846286701209,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1099435586971982e-05,
+      "loss": 0.1758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6918622371228285,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.1032506305165555e-05,
+      "loss": 0.1018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6928781875444479,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.0965722245370641e-05,
+      "loss": 0.1485,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6938941379660672,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0899084101950561e-05,
+      "loss": 0.1762,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6949100883876866,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.0832592567751555e-05,
+      "loss": 0.1402,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6959260388093061,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0766248334095505e-05,
+      "loss": 0.2278,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6969419892309255,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.0700052090772828e-05,
+      "loss": 0.1969,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6979579396525449,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.0634004526035249e-05,
+      "loss": 0.2073,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6989738900741643,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.0568106326588645e-05,
+      "loss": 0.1902,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6999898404957838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.0502358177585953e-05,
+      "loss": 0.2165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7010057909174032,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.0436760762619977e-05,
+      "loss": 0.1952,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7020217413390226,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.0371314763716347e-05,
+      "loss": 0.1422,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7030376917606421,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.0306020861326388e-05,
+      "loss": 0.0961,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7040536421822615,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.0240879734320068e-05,
+      "loss": 0.1542,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7050695926038809,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.0175892059978901e-05,
+      "loss": 0.1748,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7060855430255003,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.0111058513988958e-05,
+      "loss": 0.0819,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7071014934471198,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.0046379770433803e-05,
+      "loss": 0.1933,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7081174438687392,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.98185650178749e-06,
+      "loss": 0.1891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7091333942903586,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.917489378907591e-06,
+      "loss": 0.2102,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7101493447119781,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.853279071028212e-06,
+      "loss": 0.1714,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7111652951335975,
+      "grad_norm": 2.375,
+      "learning_rate": 9.78922624575303e-06,
+      "loss": 0.1299,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7121812455552169,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.72533156904833e-06,
+      "loss": 0.1914,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7131971959768363,
+      "grad_norm": 3.859375,
+      "learning_rate": 9.661595705236137e-06,
+      "loss": 0.2377,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7142131463984558,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.598019316987244e-06,
+      "loss": 0.1851,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7152290968200752,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.53460306531439e-06,
+      "loss": 0.2661,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7162450472416946,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.471347609565311e-06,
+      "loss": 0.1669,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7172609976633141,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.408253607415957e-06,
+      "loss": 0.2487,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7182769480849335,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.345321714863614e-06,
+      "loss": 0.186,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7192928985065529,
+      "grad_norm": 6.0625,
+      "learning_rate": 9.282552586220075e-06,
+      "loss": 0.2249,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7203088489281723,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.219946874104885e-06,
+      "loss": 0.1255,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7213247993497918,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.157505229438481e-06,
+      "loss": 0.1999,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7223407497714112,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.095228301435518e-06,
+      "loss": 0.199,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7233567001930306,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.03311673759802e-06,
+      "loss": 0.2182,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7243726506146501,
+      "grad_norm": 6.46875,
+      "learning_rate": 8.971171183708733e-06,
+      "loss": 0.1573,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7253886010362695,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.909392283824353e-06,
+      "loss": 0.2044,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7264045514578888,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.847780680268872e-06,
+      "loss": 0.11,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7274205018795082,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.786337013626853e-06,
+      "loss": 0.1897,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7284364523011277,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.725061922736799e-06,
+      "loss": 0.153,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7294524027227471,
+      "grad_norm": 1.609375,
+      "learning_rate": 8.663956044684532e-06,
+      "loss": 0.1746,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7304683531443665,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.603020014796507e-06,
+      "loss": 0.2284,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7314843035659859,
+      "grad_norm": 1.515625,
+      "learning_rate": 8.542254466633273e-06,
+      "loss": 0.1186,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7325002539876054,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.481660031982844e-06,
+      "loss": 0.1971,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7335162044092248,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.421237340854157e-06,
+      "loss": 0.196,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7345321548308442,
+      "grad_norm": 0.65234375,
+      "learning_rate": 8.360987021470479e-06,
+      "loss": 0.1724,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7355481052524637,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.300909700262929e-06,
+      "loss": 0.175,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7365640556740831,
+      "grad_norm": 3.109375,
+      "learning_rate": 8.241006001863924e-06,
+      "loss": 0.2276,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7375800060957025,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.181276549100714e-06,
+      "loss": 0.2029,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7385959565173219,
+      "grad_norm": 4.03125,
+      "learning_rate": 8.12172196298887e-06,
+      "loss": 0.175,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7396119069389414,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.062342862725878e-06,
+      "loss": 0.1662,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7406278573605608,
+      "grad_norm": 3.375,
+      "learning_rate": 8.003139865684662e-06,
+      "loss": 0.1616,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7416438077821802,
+      "grad_norm": 2.5625,
+      "learning_rate": 7.944113587407157e-06,
+      "loss": 0.2448,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7426597582037997,
+      "grad_norm": 4.125,
+      "learning_rate": 7.885264641597961e-06,
+      "loss": 0.1618,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7436757086254191,
+      "grad_norm": 3.5,
+      "learning_rate": 7.826593640117889e-06,
+      "loss": 0.1134,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7446916590470385,
+      "grad_norm": 2.6875,
+      "learning_rate": 7.76810119297767e-06,
+      "loss": 0.1795,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7457076094686579,
+      "grad_norm": 4.34375,
+      "learning_rate": 7.709787908331556e-06,
+      "loss": 0.2736,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7467235598902774,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.651654392471038e-06,
+      "loss": 0.139,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7477395103118968,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.593701249818521e-06,
+      "loss": 0.2023,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7487554607335162,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.535929082921048e-06,
+      "loss": 0.1702,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7497714111551357,
+      "grad_norm": 1.96875,
+      "learning_rate": 7.47833849244402e-06,
+      "loss": 0.1835,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7507873615767551,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.420930077164959e-06,
+      "loss": 0.1713,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7518033119983745,
+      "grad_norm": 4.46875,
+      "learning_rate": 7.363704433967311e-06,
+      "loss": 0.1906,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7528192624199939,
+      "grad_norm": 1.75,
+      "learning_rate": 7.306662157834185e-06,
+      "loss": 0.1421,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7538352128416134,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.2498038418422145e-06,
+      "loss": 0.1793,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7548511632632328,
+      "grad_norm": 2.578125,
+      "learning_rate": 7.193130077155374e-06,
+      "loss": 0.1603,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7558671136848522,
+      "grad_norm": 4.3125,
+      "learning_rate": 7.13664145301883e-06,
+      "loss": 0.2169,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7568830641064717,
+      "grad_norm": 3.078125,
+      "learning_rate": 7.0803385567528025e-06,
+      "loss": 0.1685,
+      "step": 7450
+    },
+    {
+      "epoch": 0.757899014528091,
+      "grad_norm": 3.5625,
+      "learning_rate": 7.024221973746495e-06,
+      "loss": 0.2282,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7589149649497104,
+      "grad_norm": 2.265625,
+      "learning_rate": 6.968292287451961e-06,
+      "loss": 0.1786,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7599309153713298,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.912550079378091e-06,
+      "loss": 0.1811,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7609468657929493,
+      "grad_norm": 2.328125,
+      "learning_rate": 6.856995929084506e-06,
+      "loss": 0.1747,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7619628162145687,
+      "grad_norm": 5.21875,
+      "learning_rate": 6.801630414175589e-06,
+      "loss": 0.2028,
+      "step": 7500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-7500/training_args.bin b/checkpoints/checkpoint-7500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-7500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-8000/adapter_config.json b/checkpoints/checkpoint-8000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-8000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-8000/adapter_model.safetensors b/checkpoints/checkpoint-8000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a3e74da16eac9d652cb024c1c96567e92d1eb6b2
--- /dev/null
+++ b/checkpoints/checkpoint-8000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34815b44876635d03a12d793111d30ddb5f1ba50b8380dcfca1bf6c33938f840
+size 5919456
diff --git a/checkpoints/checkpoint-8000/optimizer.pt b/checkpoints/checkpoint-8000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..756d0512f4d3fe7c92a0b851ef0cade6d0d5d29c
--- /dev/null
+++ b/checkpoints/checkpoint-8000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f76507605c949d2190ab2abb8d665fe079ec56e9d7d3261ffaa91dcc3e884b3
+size 11930938
diff --git a/checkpoints/checkpoint-8000/rng_state_0.pth b/checkpoints/checkpoint-8000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..820bae5762cf93cad79c7772fa2a022119a232d9
--- /dev/null
+++ b/checkpoints/checkpoint-8000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f2044a5599a741319339b0e82e423a0b7bec9a103bb74fa855a2b26076b2bbe
+size 15024
diff --git a/checkpoints/checkpoint-8000/rng_state_1.pth b/checkpoints/checkpoint-8000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..1cab1e079612b70518a476c16ab1414d88aff839
--- /dev/null
+++ b/checkpoints/checkpoint-8000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:274ebe0b40c3a495361afcc0dde20f0438b9b75835963738c84e733228f7478e
+size 15024
diff --git a/checkpoints/checkpoint-8000/rng_state_2.pth b/checkpoints/checkpoint-8000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..00aef9053f81e7da4386c3746977735f10ec06c9
--- /dev/null
+++ b/checkpoints/checkpoint-8000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd18781ccda8ced5e8f822443d08381a84f46d37b63c3de7993f25d8a7b0dac0
+size 15024
diff --git a/checkpoints/checkpoint-8000/rng_state_3.pth b/checkpoints/checkpoint-8000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..928e4beb3f21210417e2df6c6c1dcd68a708f0ee
--- /dev/null
+++ b/checkpoints/checkpoint-8000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a1dc1515069dbf95fe772f9cfe866820d39bf12f126251b05b193492bf8026b
+size 15024
diff --git a/checkpoints/checkpoint-8000/scheduler.pt b/checkpoints/checkpoint-8000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3e3e70b0e2f0324728f6d70898373ef2914d119a
--- /dev/null
+++ b/checkpoints/checkpoint-8000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d390c33d5dd3d95d913d2faf47a8c88adcdfd442ecba2876f37ca1daeb1d2bf4
+size 1064
diff --git a/checkpoints/checkpoint-8000/trainer_state.json b/checkpoints/checkpoint-8000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9cd92bbfdbf0eb2401ba4a0d81d37b675b3a870f
--- /dev/null
+++ b/checkpoints/checkpoint-8000/trainer_state.json
@@ -0,0 +1,5621 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.81276033729554,
+  "eval_steps": 500,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6613837244742456,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.3100382346268392e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 0.662399674895865,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.3029550230900812e-05,
+      "loss": 0.1997,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6634156253174845,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2958842573996016e-05,
+      "loss": 0.1969,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6644315757391039,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2888260110711525e-05,
+      "loss": 0.1469,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6654475261607233,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.2817803574903212e-05,
+      "loss": 0.1524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6664634765823427,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2747473699117668e-05,
+      "loss": 0.159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6674794270039622,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.267727121458458e-05,
+      "loss": 0.1999,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6684953774255816,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.2607196851209137e-05,
+      "loss": 0.2216,
+      "step": 6580
+    },
+    {
+      "epoch": 0.669511327847201,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2537251337564412e-05,
+      "loss": 0.1607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6705272782688205,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.2467435400883839e-05,
+      "loss": 0.2187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6715432286904399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.239774976705359e-05,
+      "loss": 0.1753,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6725591791120593,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2328195160605092e-05,
+      "loss": 0.194,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6735751295336787,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.225877230470743e-05,
+      "loss": 0.1485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6745910799552982,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.218948192115988e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6756070303769176,
+      "grad_norm": 3.875,
+      "learning_rate": 1.21203247303844e-05,
+      "loss": 0.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.676622980798537,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.2051301451418073e-05,
+      "loss": 0.2377,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6776389312201565,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.198241280190574e-05,
+      "loss": 0.1508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6786548816417759,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.1913659498092431e-05,
+      "loss": 0.1537,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6796708320633953,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.184504225481601e-05,
+      "loss": 0.2339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6806867824850147,
+      "grad_norm": 5.625,
+      "learning_rate": 1.177656178549966e-05,
+      "loss": 0.2102,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6817027329066342,
+      "grad_norm": 2.5,
+      "learning_rate": 1.1708218802144536e-05,
+      "loss": 0.1435,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6827186833282536,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1640014015322323e-05,
+      "loss": 0.1823,
+      "step": 6720
+    },
+    {
+      "epoch": 0.683734633749873,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1571948134167862e-05,
+      "loss": 0.1154,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6847505841714925,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1504021866371761e-05,
+      "loss": 0.2105,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6857665345931119,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.143623591817304e-05,
+      "loss": 0.1317,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6867824850147313,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.1368590994351835e-05,
+      "loss": 0.1406,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6877984354363507,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.130108779822198e-05,
+      "loss": 0.1425,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6888143858579702,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.1233727031623783e-05,
+      "loss": 0.1623,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6898303362795896,
+      "grad_norm": 4.625,
+      "learning_rate": 1.1166509394916682e-05,
+      "loss": 0.1591,
+      "step": 6790
+    },
+    {
+      "epoch": 0.690846286701209,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1099435586971982e-05,
+      "loss": 0.1758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6918622371228285,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.1032506305165555e-05,
+      "loss": 0.1018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6928781875444479,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.0965722245370641e-05,
+      "loss": 0.1485,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6938941379660672,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0899084101950561e-05,
+      "loss": 0.1762,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6949100883876866,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.0832592567751555e-05,
+      "loss": 0.1402,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6959260388093061,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0766248334095505e-05,
+      "loss": 0.2278,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6969419892309255,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.0700052090772828e-05,
+      "loss": 0.1969,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6979579396525449,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.0634004526035249e-05,
+      "loss": 0.2073,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6989738900741643,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.0568106326588645e-05,
+      "loss": 0.1902,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6999898404957838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.0502358177585953e-05,
+      "loss": 0.2165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7010057909174032,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.0436760762619977e-05,
+      "loss": 0.1952,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7020217413390226,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.0371314763716347e-05,
+      "loss": 0.1422,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7030376917606421,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.0306020861326388e-05,
+      "loss": 0.0961,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7040536421822615,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.0240879734320068e-05,
+      "loss": 0.1542,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7050695926038809,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.0175892059978901e-05,
+      "loss": 0.1748,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7060855430255003,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.0111058513988958e-05,
+      "loss": 0.0819,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7071014934471198,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.0046379770433803e-05,
+      "loss": 0.1933,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7081174438687392,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.98185650178749e-06,
+      "loss": 0.1891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7091333942903586,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.917489378907591e-06,
+      "loss": 0.2102,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7101493447119781,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.853279071028212e-06,
+      "loss": 0.1714,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7111652951335975,
+      "grad_norm": 2.375,
+      "learning_rate": 9.78922624575303e-06,
+      "loss": 0.1299,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7121812455552169,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.72533156904833e-06,
+      "loss": 0.1914,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7131971959768363,
+      "grad_norm": 3.859375,
+      "learning_rate": 9.661595705236137e-06,
+      "loss": 0.2377,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7142131463984558,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.598019316987244e-06,
+      "loss": 0.1851,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7152290968200752,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.53460306531439e-06,
+      "loss": 0.2661,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7162450472416946,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.471347609565311e-06,
+      "loss": 0.1669,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7172609976633141,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.408253607415957e-06,
+      "loss": 0.2487,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7182769480849335,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.345321714863614e-06,
+      "loss": 0.186,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7192928985065529,
+      "grad_norm": 6.0625,
+      "learning_rate": 9.282552586220075e-06,
+      "loss": 0.2249,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7203088489281723,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.219946874104885e-06,
+      "loss": 0.1255,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7213247993497918,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.157505229438481e-06,
+      "loss": 0.1999,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7223407497714112,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.095228301435518e-06,
+      "loss": 0.199,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7233567001930306,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.03311673759802e-06,
+      "loss": 0.2182,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7243726506146501,
+      "grad_norm": 6.46875,
+      "learning_rate": 8.971171183708733e-06,
+      "loss": 0.1573,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7253886010362695,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.909392283824353e-06,
+      "loss": 0.2044,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7264045514578888,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.847780680268872e-06,
+      "loss": 0.11,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7274205018795082,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.786337013626853e-06,
+      "loss": 0.1897,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7284364523011277,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.725061922736799e-06,
+      "loss": 0.153,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7294524027227471,
+      "grad_norm": 1.609375,
+      "learning_rate": 8.663956044684532e-06,
+      "loss": 0.1746,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7304683531443665,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.603020014796507e-06,
+      "loss": 0.2284,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7314843035659859,
+      "grad_norm": 1.515625,
+      "learning_rate": 8.542254466633273e-06,
+      "loss": 0.1186,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7325002539876054,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.481660031982844e-06,
+      "loss": 0.1971,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7335162044092248,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.421237340854157e-06,
+      "loss": 0.196,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7345321548308442,
+      "grad_norm": 0.65234375,
+      "learning_rate": 8.360987021470479e-06,
+      "loss": 0.1724,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7355481052524637,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.300909700262929e-06,
+      "loss": 0.175,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7365640556740831,
+      "grad_norm": 3.109375,
+      "learning_rate": 8.241006001863924e-06,
+      "loss": 0.2276,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7375800060957025,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.181276549100714e-06,
+      "loss": 0.2029,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7385959565173219,
+      "grad_norm": 4.03125,
+      "learning_rate": 8.12172196298887e-06,
+      "loss": 0.175,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7396119069389414,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.062342862725878e-06,
+      "loss": 0.1662,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7406278573605608,
+      "grad_norm": 3.375,
+      "learning_rate": 8.003139865684662e-06,
+      "loss": 0.1616,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7416438077821802,
+      "grad_norm": 2.5625,
+      "learning_rate": 7.944113587407157e-06,
+      "loss": 0.2448,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7426597582037997,
+      "grad_norm": 4.125,
+      "learning_rate": 7.885264641597961e-06,
+      "loss": 0.1618,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7436757086254191,
+      "grad_norm": 3.5,
+      "learning_rate": 7.826593640117889e-06,
+      "loss": 0.1134,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7446916590470385,
+      "grad_norm": 2.6875,
+      "learning_rate": 7.76810119297767e-06,
+      "loss": 0.1795,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7457076094686579,
+      "grad_norm": 4.34375,
+      "learning_rate": 7.709787908331556e-06,
+      "loss": 0.2736,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7467235598902774,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.651654392471038e-06,
+      "loss": 0.139,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7477395103118968,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.593701249818521e-06,
+      "loss": 0.2023,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7487554607335162,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.535929082921048e-06,
+      "loss": 0.1702,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7497714111551357,
+      "grad_norm": 1.96875,
+      "learning_rate": 7.47833849244402e-06,
+      "loss": 0.1835,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7507873615767551,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.420930077164959e-06,
+      "loss": 0.1713,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7518033119983745,
+      "grad_norm": 4.46875,
+      "learning_rate": 7.363704433967311e-06,
+      "loss": 0.1906,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7528192624199939,
+      "grad_norm": 1.75,
+      "learning_rate": 7.306662157834185e-06,
+      "loss": 0.1421,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7538352128416134,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.2498038418422145e-06,
+      "loss": 0.1793,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7548511632632328,
+      "grad_norm": 2.578125,
+      "learning_rate": 7.193130077155374e-06,
+      "loss": 0.1603,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7558671136848522,
+      "grad_norm": 4.3125,
+      "learning_rate": 7.13664145301883e-06,
+      "loss": 0.2169,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7568830641064717,
+      "grad_norm": 3.078125,
+      "learning_rate": 7.0803385567528025e-06,
+      "loss": 0.1685,
+      "step": 7450
+    },
+    {
+      "epoch": 0.757899014528091,
+      "grad_norm": 3.5625,
+      "learning_rate": 7.024221973746495e-06,
+      "loss": 0.2282,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7589149649497104,
+      "grad_norm": 2.265625,
+      "learning_rate": 6.968292287451961e-06,
+      "loss": 0.1786,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7599309153713298,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.912550079378091e-06,
+      "loss": 0.1811,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7609468657929493,
+      "grad_norm": 2.328125,
+      "learning_rate": 6.856995929084506e-06,
+      "loss": 0.1747,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7619628162145687,
+      "grad_norm": 5.21875,
+      "learning_rate": 6.801630414175589e-06,
+      "loss": 0.2028,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7629787666361881,
+      "grad_norm": 3.78125,
+      "learning_rate": 6.746454110294451e-06,
+      "loss": 0.2255,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7639947170578075,
+      "grad_norm": 1.625,
+      "learning_rate": 6.691467591116931e-06,
+      "loss": 0.1604,
+      "step": 7520
+    },
+    {
+      "epoch": 0.765010667479427,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.6366714283456755e-06,
+      "loss": 0.2559,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7660266179010464,
+      "grad_norm": 4.59375,
+      "learning_rate": 6.582066191704142e-06,
+      "loss": 0.2034,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7670425683226658,
+      "grad_norm": 1.578125,
+      "learning_rate": 6.527652448930724e-06,
+      "loss": 0.148,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7680585187442853,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.4734307657728e-06,
+      "loss": 0.1811,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7690744691659047,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.419401705980924e-06,
+      "loss": 0.1407,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7700904195875241,
+      "grad_norm": 2.25,
+      "learning_rate": 6.365565831302869e-06,
+      "loss": 0.1893,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7711063700091435,
+      "grad_norm": 1.625,
+      "learning_rate": 6.311923701477854e-06,
+      "loss": 0.1835,
+      "step": 7590
+    },
+    {
+      "epoch": 0.772122320430763,
+      "grad_norm": 2.375,
+      "learning_rate": 6.258475874230713e-06,
+      "loss": 0.1579,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7731382708523824,
+      "grad_norm": 4.5,
+      "learning_rate": 6.205222905266067e-06,
+      "loss": 0.1794,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7741542212740018,
+      "grad_norm": 4.25,
+      "learning_rate": 6.152165348262598e-06,
+      "loss": 0.1477,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7751701716956213,
+      "grad_norm": 1.9765625,
+      "learning_rate": 6.0993037548672246e-06,
+      "loss": 0.2396,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7761861221172407,
+      "grad_norm": 2.671875,
+      "learning_rate": 6.046638674689454e-06,
+      "loss": 0.1717,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7772020725388601,
+      "grad_norm": 3.671875,
+      "learning_rate": 5.994170655295567e-06,
+      "loss": 0.2646,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7782180229604795,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.9419002422030106e-06,
+      "loss": 0.1553,
+      "step": 7660
+    },
+    {
+      "epoch": 0.779233973382099,
+      "grad_norm": 3.734375,
+      "learning_rate": 5.889827978874665e-06,
+      "loss": 0.1854,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7802499238037184,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.837954406713245e-06,
+      "loss": 0.1857,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7812658742253378,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.786280065055619e-06,
+      "loss": 0.1797,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7822818246469573,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.734805491167244e-06,
+      "loss": 0.1488,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7832977750685767,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.683531220236576e-06,
+      "loss": 0.1688,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.632457785369455e-06,
+      "loss": 0.1503,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7853296759118155,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.581585717583637e-06,
+      "loss": 0.1658,
+      "step": 7730
+    },
+    {
+      "epoch": 0.786345626333435,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.530915545803209e-06,
+      "loss": 0.2112,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7873615767550544,
+      "grad_norm": 4.1875,
+      "learning_rate": 5.480447796853141e-06,
+      "loss": 0.165,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7883775271766738,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.430182995453756e-06,
+      "loss": 0.1499,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7893934775982933,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.380121664215329e-06,
+      "loss": 0.1559,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7904094280199127,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.330264323632611e-06,
+      "loss": 0.2098,
+      "step": 7780
+    },
+    {
+      "epoch": 0.791425378441532,
+      "grad_norm": 4.65625,
+      "learning_rate": 5.280611492079449e-06,
+      "loss": 0.1776,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7924413288631514,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.231163685803361e-06,
+      "loss": 0.1497,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7934572792847709,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.181921418920191e-06,
+      "loss": 0.12,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7944732297063903,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.13288520340878e-06,
+      "loss": 0.1981,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7954891801280097,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.084055549105596e-06,
+      "loss": 0.1389,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7965051305496291,
+      "grad_norm": 2.796875,
+      "learning_rate": 5.035432963699479e-06,
+      "loss": 0.2293,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7975210809712486,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.98701795272635e-06,
+      "loss": 0.1618,
+      "step": 7850
+    },
+    {
+      "epoch": 0.798537031392868,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.938811019563938e-06,
+      "loss": 0.1755,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7995529818144874,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.8908126654265475e-06,
+      "loss": 0.1565,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8005689322361069,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.843023389359885e-06,
+      "loss": 0.2176,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8015848826577263,
+      "grad_norm": 2.625,
+      "learning_rate": 4.79544368823581e-06,
+      "loss": 0.2013,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8026008330793457,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.748074056747234e-06,
+      "loss": 0.1246,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8036167835009651,
+      "grad_norm": 3.5,
+      "learning_rate": 4.700914987402919e-06,
+      "loss": 0.1638,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8046327339225846,
+      "grad_norm": 3.4375,
+      "learning_rate": 4.6539669705223916e-06,
+      "loss": 0.2213,
+      "step": 7920
+    },
+    {
+      "epoch": 0.805648684344204,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.607230494230849e-06,
+      "loss": 0.1822,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8066646347658234,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.560706044454047e-06,
+      "loss": 0.1763,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8076805851874429,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.514394104913291e-06,
+      "loss": 0.234,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8086965356090623,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.468295157120372e-06,
+      "loss": 0.1939,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8097124860306817,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.422409680372594e-06,
+      "loss": 0.174,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8107284364523011,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.3767381517477505e-06,
+      "loss": 0.2375,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8117443868739206,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.331281046099203e-06,
+      "loss": 0.2076,
+      "step": 7990
+    },
+    {
+      "epoch": 0.81276033729554,
+      "grad_norm": 6.0625,
+      "learning_rate": 4.286038836050929e-06,
+      "loss": 0.2504,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-8000/training_args.bin b/checkpoints/checkpoint-8000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-8000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-8500/adapter_config.json b/checkpoints/checkpoint-8500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-8500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-8500/adapter_model.safetensors b/checkpoints/checkpoint-8500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c4142c2f632b19891dd52f1c81cfe510afb044ab
--- /dev/null
+++ b/checkpoints/checkpoint-8500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28ac81ab3c97c69dfd1334375e0193bcbeec6d6567bc44b70349e272cd340ee0
+size 5919456
diff --git a/checkpoints/checkpoint-8500/optimizer.pt b/checkpoints/checkpoint-8500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0eb48e2be9f73c79b44660a595421fd4cb1d762c
--- /dev/null
+++ b/checkpoints/checkpoint-8500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a97970da7a377ddc03d41ac023f8d6e5a218717457c493a71d03647daee1d293
+size 11930938
diff --git a/checkpoints/checkpoint-8500/rng_state_0.pth b/checkpoints/checkpoint-8500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..de8f82b7c4feb8b730311d766bf44d581384eae7
--- /dev/null
+++ b/checkpoints/checkpoint-8500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a353686e04717107f6f8ef46d31bd168a04a71491dff1b49a95dd841cae20dee
+size 15024
diff --git a/checkpoints/checkpoint-8500/rng_state_1.pth b/checkpoints/checkpoint-8500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..58bca56d46a97b8f2ee454510a5f93c7f3bb7410
--- /dev/null
+++ b/checkpoints/checkpoint-8500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59a024dac60d84768b968adca05416d784c0295ae139bead6f2816285ba6b01b
+size 15024
diff --git a/checkpoints/checkpoint-8500/rng_state_2.pth b/checkpoints/checkpoint-8500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..9748eeb88596cf9f27c2821bec97e56aa5f2ccfc
--- /dev/null
+++ b/checkpoints/checkpoint-8500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f28ab7ae50115d00765b2d4d2188a1d88b5bcb9be07f45fe7e78cb4752256bd
+size 15024
diff --git a/checkpoints/checkpoint-8500/rng_state_3.pth b/checkpoints/checkpoint-8500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..354e6dd8952c6a6e554d8c56f7f0daa728e13f22
--- /dev/null
+++ b/checkpoints/checkpoint-8500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce60533c28c44e2a4955c454dfc9eac41778818fa945283e28161cac010c0143
+size 15024
diff --git a/checkpoints/checkpoint-8500/scheduler.pt b/checkpoints/checkpoint-8500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f22ac3e321eef132a6a05d855a039dde00c7589a
--- /dev/null
+++ b/checkpoints/checkpoint-8500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d013f7dc21c771c94b398b7ee47f1c3b60572d9f3b478e50ee4f2b58d385070d
+size 1064
diff --git a/checkpoints/checkpoint-8500/trainer_state.json b/checkpoints/checkpoint-8500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..3ef48260e073af534f11c2fb65f8473ce1a4a401
--- /dev/null
+++ b/checkpoints/checkpoint-8500/trainer_state.json
@@ -0,0 +1,5971 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8635578583765112,
+  "eval_steps": 500,
+  "global_step": 8500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6613837244742456,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.3100382346268392e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 0.662399674895865,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.3029550230900812e-05,
+      "loss": 0.1997,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6634156253174845,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2958842573996016e-05,
+      "loss": 0.1969,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6644315757391039,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2888260110711525e-05,
+      "loss": 0.1469,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6654475261607233,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.2817803574903212e-05,
+      "loss": 0.1524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6664634765823427,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2747473699117668e-05,
+      "loss": 0.159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6674794270039622,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.267727121458458e-05,
+      "loss": 0.1999,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6684953774255816,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.2607196851209137e-05,
+      "loss": 0.2216,
+      "step": 6580
+    },
+    {
+      "epoch": 0.669511327847201,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2537251337564412e-05,
+      "loss": 0.1607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6705272782688205,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.2467435400883839e-05,
+      "loss": 0.2187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6715432286904399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.239774976705359e-05,
+      "loss": 0.1753,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6725591791120593,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2328195160605092e-05,
+      "loss": 0.194,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6735751295336787,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.225877230470743e-05,
+      "loss": 0.1485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6745910799552982,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.218948192115988e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6756070303769176,
+      "grad_norm": 3.875,
+      "learning_rate": 1.21203247303844e-05,
+      "loss": 0.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.676622980798537,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.2051301451418073e-05,
+      "loss": 0.2377,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6776389312201565,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.198241280190574e-05,
+      "loss": 0.1508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6786548816417759,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.1913659498092431e-05,
+      "loss": 0.1537,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6796708320633953,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.184504225481601e-05,
+      "loss": 0.2339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6806867824850147,
+      "grad_norm": 5.625,
+      "learning_rate": 1.177656178549966e-05,
+      "loss": 0.2102,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6817027329066342,
+      "grad_norm": 2.5,
+      "learning_rate": 1.1708218802144536e-05,
+      "loss": 0.1435,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6827186833282536,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1640014015322323e-05,
+      "loss": 0.1823,
+      "step": 6720
+    },
+    {
+      "epoch": 0.683734633749873,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1571948134167862e-05,
+      "loss": 0.1154,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6847505841714925,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1504021866371761e-05,
+      "loss": 0.2105,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6857665345931119,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.143623591817304e-05,
+      "loss": 0.1317,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6867824850147313,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.1368590994351835e-05,
+      "loss": 0.1406,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6877984354363507,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.130108779822198e-05,
+      "loss": 0.1425,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6888143858579702,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.1233727031623783e-05,
+      "loss": 0.1623,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6898303362795896,
+      "grad_norm": 4.625,
+      "learning_rate": 1.1166509394916682e-05,
+      "loss": 0.1591,
+      "step": 6790
+    },
+    {
+      "epoch": 0.690846286701209,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1099435586971982e-05,
+      "loss": 0.1758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6918622371228285,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.1032506305165555e-05,
+      "loss": 0.1018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6928781875444479,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.0965722245370641e-05,
+      "loss": 0.1485,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6938941379660672,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0899084101950561e-05,
+      "loss": 0.1762,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6949100883876866,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.0832592567751555e-05,
+      "loss": 0.1402,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6959260388093061,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0766248334095505e-05,
+      "loss": 0.2278,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6969419892309255,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.0700052090772828e-05,
+      "loss": 0.1969,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6979579396525449,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.0634004526035249e-05,
+      "loss": 0.2073,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6989738900741643,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.0568106326588645e-05,
+      "loss": 0.1902,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6999898404957838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.0502358177585953e-05,
+      "loss": 0.2165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7010057909174032,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.0436760762619977e-05,
+      "loss": 0.1952,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7020217413390226,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.0371314763716347e-05,
+      "loss": 0.1422,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7030376917606421,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.0306020861326388e-05,
+      "loss": 0.0961,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7040536421822615,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.0240879734320068e-05,
+      "loss": 0.1542,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7050695926038809,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.0175892059978901e-05,
+      "loss": 0.1748,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7060855430255003,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.0111058513988958e-05,
+      "loss": 0.0819,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7071014934471198,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.0046379770433803e-05,
+      "loss": 0.1933,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7081174438687392,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.98185650178749e-06,
+      "loss": 0.1891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7091333942903586,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.917489378907591e-06,
+      "loss": 0.2102,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7101493447119781,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.853279071028212e-06,
+      "loss": 0.1714,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7111652951335975,
+      "grad_norm": 2.375,
+      "learning_rate": 9.78922624575303e-06,
+      "loss": 0.1299,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7121812455552169,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.72533156904833e-06,
+      "loss": 0.1914,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7131971959768363,
+      "grad_norm": 3.859375,
+      "learning_rate": 9.661595705236137e-06,
+      "loss": 0.2377,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7142131463984558,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.598019316987244e-06,
+      "loss": 0.1851,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7152290968200752,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.53460306531439e-06,
+      "loss": 0.2661,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7162450472416946,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.471347609565311e-06,
+      "loss": 0.1669,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7172609976633141,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.408253607415957e-06,
+      "loss": 0.2487,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7182769480849335,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.345321714863614e-06,
+      "loss": 0.186,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7192928985065529,
+      "grad_norm": 6.0625,
+      "learning_rate": 9.282552586220075e-06,
+      "loss": 0.2249,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7203088489281723,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.219946874104885e-06,
+      "loss": 0.1255,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7213247993497918,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.157505229438481e-06,
+      "loss": 0.1999,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7223407497714112,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.095228301435518e-06,
+      "loss": 0.199,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7233567001930306,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.03311673759802e-06,
+      "loss": 0.2182,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7243726506146501,
+      "grad_norm": 6.46875,
+      "learning_rate": 8.971171183708733e-06,
+      "loss": 0.1573,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7253886010362695,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.909392283824353e-06,
+      "loss": 0.2044,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7264045514578888,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.847780680268872e-06,
+      "loss": 0.11,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7274205018795082,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.786337013626853e-06,
+      "loss": 0.1897,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7284364523011277,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.725061922736799e-06,
+      "loss": 0.153,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7294524027227471,
+      "grad_norm": 1.609375,
+      "learning_rate": 8.663956044684532e-06,
+      "loss": 0.1746,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7304683531443665,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.603020014796507e-06,
+      "loss": 0.2284,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7314843035659859,
+      "grad_norm": 1.515625,
+      "learning_rate": 8.542254466633273e-06,
+      "loss": 0.1186,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7325002539876054,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.481660031982844e-06,
+      "loss": 0.1971,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7335162044092248,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.421237340854157e-06,
+      "loss": 0.196,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7345321548308442,
+      "grad_norm": 0.65234375,
+      "learning_rate": 8.360987021470479e-06,
+      "loss": 0.1724,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7355481052524637,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.300909700262929e-06,
+      "loss": 0.175,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7365640556740831,
+      "grad_norm": 3.109375,
+      "learning_rate": 8.241006001863924e-06,
+      "loss": 0.2276,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7375800060957025,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.181276549100714e-06,
+      "loss": 0.2029,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7385959565173219,
+      "grad_norm": 4.03125,
+      "learning_rate": 8.12172196298887e-06,
+      "loss": 0.175,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7396119069389414,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.062342862725878e-06,
+      "loss": 0.1662,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7406278573605608,
+      "grad_norm": 3.375,
+      "learning_rate": 8.003139865684662e-06,
+      "loss": 0.1616,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7416438077821802,
+      "grad_norm": 2.5625,
+      "learning_rate": 7.944113587407157e-06,
+      "loss": 0.2448,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7426597582037997,
+      "grad_norm": 4.125,
+      "learning_rate": 7.885264641597961e-06,
+      "loss": 0.1618,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7436757086254191,
+      "grad_norm": 3.5,
+      "learning_rate": 7.826593640117889e-06,
+      "loss": 0.1134,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7446916590470385,
+      "grad_norm": 2.6875,
+      "learning_rate": 7.76810119297767e-06,
+      "loss": 0.1795,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7457076094686579,
+      "grad_norm": 4.34375,
+      "learning_rate": 7.709787908331556e-06,
+      "loss": 0.2736,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7467235598902774,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.651654392471038e-06,
+      "loss": 0.139,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7477395103118968,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.593701249818521e-06,
+      "loss": 0.2023,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7487554607335162,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.535929082921048e-06,
+      "loss": 0.1702,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7497714111551357,
+      "grad_norm": 1.96875,
+      "learning_rate": 7.47833849244402e-06,
+      "loss": 0.1835,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7507873615767551,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.420930077164959e-06,
+      "loss": 0.1713,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7518033119983745,
+      "grad_norm": 4.46875,
+      "learning_rate": 7.363704433967311e-06,
+      "loss": 0.1906,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7528192624199939,
+      "grad_norm": 1.75,
+      "learning_rate": 7.306662157834185e-06,
+      "loss": 0.1421,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7538352128416134,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.2498038418422145e-06,
+      "loss": 0.1793,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7548511632632328,
+      "grad_norm": 2.578125,
+      "learning_rate": 7.193130077155374e-06,
+      "loss": 0.1603,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7558671136848522,
+      "grad_norm": 4.3125,
+      "learning_rate": 7.13664145301883e-06,
+      "loss": 0.2169,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7568830641064717,
+      "grad_norm": 3.078125,
+      "learning_rate": 7.0803385567528025e-06,
+      "loss": 0.1685,
+      "step": 7450
+    },
+    {
+      "epoch": 0.757899014528091,
+      "grad_norm": 3.5625,
+      "learning_rate": 7.024221973746495e-06,
+      "loss": 0.2282,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7589149649497104,
+      "grad_norm": 2.265625,
+      "learning_rate": 6.968292287451961e-06,
+      "loss": 0.1786,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7599309153713298,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.912550079378091e-06,
+      "loss": 0.1811,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7609468657929493,
+      "grad_norm": 2.328125,
+      "learning_rate": 6.856995929084506e-06,
+      "loss": 0.1747,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7619628162145687,
+      "grad_norm": 5.21875,
+      "learning_rate": 6.801630414175589e-06,
+      "loss": 0.2028,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7629787666361881,
+      "grad_norm": 3.78125,
+      "learning_rate": 6.746454110294451e-06,
+      "loss": 0.2255,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7639947170578075,
+      "grad_norm": 1.625,
+      "learning_rate": 6.691467591116931e-06,
+      "loss": 0.1604,
+      "step": 7520
+    },
+    {
+      "epoch": 0.765010667479427,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.6366714283456755e-06,
+      "loss": 0.2559,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7660266179010464,
+      "grad_norm": 4.59375,
+      "learning_rate": 6.582066191704142e-06,
+      "loss": 0.2034,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7670425683226658,
+      "grad_norm": 1.578125,
+      "learning_rate": 6.527652448930724e-06,
+      "loss": 0.148,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7680585187442853,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.4734307657728e-06,
+      "loss": 0.1811,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7690744691659047,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.419401705980924e-06,
+      "loss": 0.1407,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7700904195875241,
+      "grad_norm": 2.25,
+      "learning_rate": 6.365565831302869e-06,
+      "loss": 0.1893,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7711063700091435,
+      "grad_norm": 1.625,
+      "learning_rate": 6.311923701477854e-06,
+      "loss": 0.1835,
+      "step": 7590
+    },
+    {
+      "epoch": 0.772122320430763,
+      "grad_norm": 2.375,
+      "learning_rate": 6.258475874230713e-06,
+      "loss": 0.1579,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7731382708523824,
+      "grad_norm": 4.5,
+      "learning_rate": 6.205222905266067e-06,
+      "loss": 0.1794,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7741542212740018,
+      "grad_norm": 4.25,
+      "learning_rate": 6.152165348262598e-06,
+      "loss": 0.1477,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7751701716956213,
+      "grad_norm": 1.9765625,
+      "learning_rate": 6.0993037548672246e-06,
+      "loss": 0.2396,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7761861221172407,
+      "grad_norm": 2.671875,
+      "learning_rate": 6.046638674689454e-06,
+      "loss": 0.1717,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7772020725388601,
+      "grad_norm": 3.671875,
+      "learning_rate": 5.994170655295567e-06,
+      "loss": 0.2646,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7782180229604795,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.9419002422030106e-06,
+      "loss": 0.1553,
+      "step": 7660
+    },
+    {
+      "epoch": 0.779233973382099,
+      "grad_norm": 3.734375,
+      "learning_rate": 5.889827978874665e-06,
+      "loss": 0.1854,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7802499238037184,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.837954406713245e-06,
+      "loss": 0.1857,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7812658742253378,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.786280065055619e-06,
+      "loss": 0.1797,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7822818246469573,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.734805491167244e-06,
+      "loss": 0.1488,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7832977750685767,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.683531220236576e-06,
+      "loss": 0.1688,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.632457785369455e-06,
+      "loss": 0.1503,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7853296759118155,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.581585717583637e-06,
+      "loss": 0.1658,
+      "step": 7730
+    },
+    {
+      "epoch": 0.786345626333435,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.530915545803209e-06,
+      "loss": 0.2112,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7873615767550544,
+      "grad_norm": 4.1875,
+      "learning_rate": 5.480447796853141e-06,
+      "loss": 0.165,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7883775271766738,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.430182995453756e-06,
+      "loss": 0.1499,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7893934775982933,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.380121664215329e-06,
+      "loss": 0.1559,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7904094280199127,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.330264323632611e-06,
+      "loss": 0.2098,
+      "step": 7780
+    },
+    {
+      "epoch": 0.791425378441532,
+      "grad_norm": 4.65625,
+      "learning_rate": 5.280611492079449e-06,
+      "loss": 0.1776,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7924413288631514,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.231163685803361e-06,
+      "loss": 0.1497,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7934572792847709,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.181921418920191e-06,
+      "loss": 0.12,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7944732297063903,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.13288520340878e-06,
+      "loss": 0.1981,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7954891801280097,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.084055549105596e-06,
+      "loss": 0.1389,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7965051305496291,
+      "grad_norm": 2.796875,
+      "learning_rate": 5.035432963699479e-06,
+      "loss": 0.2293,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7975210809712486,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.98701795272635e-06,
+      "loss": 0.1618,
+      "step": 7850
+    },
+    {
+      "epoch": 0.798537031392868,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.938811019563938e-06,
+      "loss": 0.1755,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7995529818144874,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.8908126654265475e-06,
+      "loss": 0.1565,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8005689322361069,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.843023389359885e-06,
+      "loss": 0.2176,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8015848826577263,
+      "grad_norm": 2.625,
+      "learning_rate": 4.79544368823581e-06,
+      "loss": 0.2013,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8026008330793457,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.748074056747234e-06,
+      "loss": 0.1246,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8036167835009651,
+      "grad_norm": 3.5,
+      "learning_rate": 4.700914987402919e-06,
+      "loss": 0.1638,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8046327339225846,
+      "grad_norm": 3.4375,
+      "learning_rate": 4.6539669705223916e-06,
+      "loss": 0.2213,
+      "step": 7920
+    },
+    {
+      "epoch": 0.805648684344204,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.607230494230849e-06,
+      "loss": 0.1822,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8066646347658234,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.560706044454047e-06,
+      "loss": 0.1763,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8076805851874429,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.514394104913291e-06,
+      "loss": 0.234,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8086965356090623,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.468295157120372e-06,
+      "loss": 0.1939,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8097124860306817,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.422409680372594e-06,
+      "loss": 0.174,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8107284364523011,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.3767381517477505e-06,
+      "loss": 0.2375,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8117443868739206,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.331281046099203e-06,
+      "loss": 0.2076,
+      "step": 7990
+    },
+    {
+      "epoch": 0.81276033729554,
+      "grad_norm": 6.0625,
+      "learning_rate": 4.286038836050929e-06,
+      "loss": 0.2504,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8137762877171594,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.241011991992586e-06,
+      "loss": 0.2102,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8147922381387789,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.1962009820746635e-06,
+      "loss": 0.1846,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8158081885603983,
+      "grad_norm": 1.875,
+      "learning_rate": 4.15160627220357e-06,
+      "loss": 0.1741,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8168241389820177,
+      "grad_norm": 5.5625,
+      "learning_rate": 4.107228326036838e-06,
+      "loss": 0.2078,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8178400894036371,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.063067604978252e-06,
+      "loss": 0.212,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8188560398252566,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.019124568173094e-06,
+      "loss": 0.1831,
+      "step": 8060
+    },
+    {
+      "epoch": 0.819871990246876,
+      "grad_norm": 6.625,
+      "learning_rate": 3.975399672503341e-06,
+      "loss": 0.2196,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8208879406684954,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.931893372582943e-06,
+      "loss": 0.2002,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8219038910901149,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.888606120753047e-06,
+      "loss": 0.2138,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8229198415117343,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.845538367077362e-06,
+      "loss": 0.2593,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8239357919333536,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.8026905593374213e-06,
+      "loss": 0.2062,
+      "step": 8110
+    },
+    {
+      "epoch": 0.824951742354973,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.760063143027945e-06,
+      "loss": 0.1343,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8259676927765925,
+      "grad_norm": 1.984375,
+      "learning_rate": 3.7176565613522313e-06,
+      "loss": 0.2494,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8269836431982119,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.675471255217516e-06,
+      "loss": 0.1502,
+      "step": 8140
+    },
+    {
+      "epoch": 0.8279995936198313,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.6335076632304175e-06,
+      "loss": 0.1256,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8290155440414507,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.5917662216923332e-06,
+      "loss": 0.1709,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8300314944630702,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.550247364594958e-06,
+      "loss": 0.1881,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8310474448846896,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.508951523615725e-06,
+      "loss": 0.1998,
+      "step": 8180
+    },
+    {
+      "epoch": 0.832063395306309,
+      "grad_norm": 2.40625,
+      "learning_rate": 3.467879128113352e-06,
+      "loss": 0.2429,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8330793457279285,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.427030605123352e-06,
+      "loss": 0.1942,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8340952961495479,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.3864063793536043e-06,
+      "loss": 0.1898,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8351112465711673,
+      "grad_norm": 5.375,
+      "learning_rate": 3.3460068731799577e-06,
+      "loss": 0.1919,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8361271969927867,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.3058325066417818e-06,
+      "loss": 0.1516,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8371431474144062,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.26588369743768e-06,
+      "loss": 0.1068,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8381590978360256,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.2261608609210653e-06,
+      "loss": 0.1203,
+      "step": 8250
+    },
+    {
+      "epoch": 0.839175048257645,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.186664410095913e-06,
+      "loss": 0.2172,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8401909986792645,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.1473947556124093e-06,
+      "loss": 0.1249,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8412069491008839,
+      "grad_norm": 2.484375,
+      "learning_rate": 3.1083523057627213e-06,
+      "loss": 0.1744,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8422228995225033,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.0695374664767353e-06,
+      "loss": 0.1772,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8432388499441227,
+      "grad_norm": 0.59375,
+      "learning_rate": 3.0309506413178397e-06,
+      "loss": 0.2302,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8442548003657422,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.9925922314787136e-06,
+      "loss": 0.1635,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8452707507873616,
+      "grad_norm": 2.34375,
+      "learning_rate": 2.954462635777194e-06,
+      "loss": 0.1573,
+      "step": 8320
+    },
+    {
+      "epoch": 0.846286701208981,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.916562250652083e-06,
+      "loss": 0.1608,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8473026516306005,
+      "grad_norm": 4.125,
+      "learning_rate": 2.878891470159048e-06,
+      "loss": 0.184,
+      "step": 8340
+    },
+    {
+      "epoch": 0.8483186020522199,
+      "grad_norm": 2.515625,
+      "learning_rate": 2.8414506859665514e-06,
+      "loss": 0.2141,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8493345524738393,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8042402873517197e-06,
+      "loss": 0.1729,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8503505028954587,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.76726066119635e-06,
+      "loss": 0.2252,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8513664533170782,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.730512191982845e-06,
+      "loss": 0.1644,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8523824037386976,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.693995261790261e-06,
+      "loss": 0.1822,
+      "step": 8390
+    },
+    {
+      "epoch": 0.853398354160317,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.657710250290285e-06,
+      "loss": 0.2068,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8544143045819365,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.621657534743327e-06,
+      "loss": 0.1224,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8554302550035559,
+      "grad_norm": 3.421875,
+      "learning_rate": 2.5858374899945804e-06,
+      "loss": 0.179,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8564462054251752,
+      "grad_norm": 3.484375,
+      "learning_rate": 2.550250488470135e-06,
+      "loss": 0.1873,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8574621558467946,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.5148969001730806e-06,
+      "loss": 0.1799,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8584781062684141,
+      "grad_norm": 1.375,
+      "learning_rate": 2.4797770926796858e-06,
+      "loss": 0.176,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8594940566900335,
+      "grad_norm": 1.8984375,
+      "learning_rate": 2.444891431135571e-06,
+      "loss": 0.1664,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8605100071116529,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.4102402782518936e-06,
+      "loss": 0.1512,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8615259575332723,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.3758239943016096e-06,
+      "loss": 0.1629,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8625419079548918,
+      "grad_norm": 5.3125,
+      "learning_rate": 2.3416429371157013e-06,
+      "loss": 0.2099,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8635578583765112,
+      "grad_norm": 5.9375,
+      "learning_rate": 2.307697462079464e-06,
+      "loss": 0.2221,
+      "step": 8500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-8500/training_args.bin b/checkpoints/checkpoint-8500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-8500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-9000/adapter_config.json b/checkpoints/checkpoint-9000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-9000/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-9000/adapter_model.safetensors b/checkpoints/checkpoint-9000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8527f2c40f383b1cf651d36a2984f5ff825b8ae2
--- /dev/null
+++ b/checkpoints/checkpoint-9000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33bdc4e327ac26650f40cb5930deff06b7f48707f893ee77c1175a4a65f6909b
+size 5919456
diff --git a/checkpoints/checkpoint-9000/optimizer.pt b/checkpoints/checkpoint-9000/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d6bc1be7888a16d8c4b7f04fab92dd9b007b67cc
--- /dev/null
+++ b/checkpoints/checkpoint-9000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0638566e76474065be273b543ece321c7e8e004ee51a20c1a1d52b2838e78bb6
+size 11930938
diff --git a/checkpoints/checkpoint-9000/rng_state_0.pth b/checkpoints/checkpoint-9000/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2f6ccedf895c2d6ac8b9d081077455c1e01eb40b
--- /dev/null
+++ b/checkpoints/checkpoint-9000/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9ca1c49d5b46a7f77afead045cfb6ddf181a9c78168c26c7eef763523361557
+size 15024
diff --git a/checkpoints/checkpoint-9000/rng_state_1.pth b/checkpoints/checkpoint-9000/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..a9526156d7a2b5afe37477f57c9c8c5878b57db6
--- /dev/null
+++ b/checkpoints/checkpoint-9000/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:321d28c2b0ca1e12bac7c5601220acbf51dba57cb5c81444c67560204eeb23e8
+size 15024
diff --git a/checkpoints/checkpoint-9000/rng_state_2.pth b/checkpoints/checkpoint-9000/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..cf5c72ead4046664eac6220c25e849e4fd4ae199
--- /dev/null
+++ b/checkpoints/checkpoint-9000/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dbe2075823c34d4e2ddbb8628dd2dfb1329f66b7c4225949042b1ffb085b4c4
+size 15024
diff --git a/checkpoints/checkpoint-9000/rng_state_3.pth b/checkpoints/checkpoint-9000/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..83bd913f93637b92c9974abbc175838c2fc91877
--- /dev/null
+++ b/checkpoints/checkpoint-9000/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94cf730a64bd09032ea105de0e172f948a4a2bd002a9ce85d137729cfbbd0578
+size 15024
diff --git a/checkpoints/checkpoint-9000/scheduler.pt b/checkpoints/checkpoint-9000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..581c38180d342706ff7b6c9ed5c479819b60323e
--- /dev/null
+++ b/checkpoints/checkpoint-9000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:777ecd9b7cc3e4361056b976c635577f61b77b3bbef2bb60189849a484abfe13
+size 1064
diff --git a/checkpoints/checkpoint-9000/trainer_state.json b/checkpoints/checkpoint-9000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef0193a85dc4d73484dbcd0610b17b8f9fd78a14
--- /dev/null
+++ b/checkpoints/checkpoint-9000/trainer_state.json
@@ -0,0 +1,6321 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9143553794574825,
+  "eval_steps": 500,
+  "global_step": 9000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6613837244742456,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.3100382346268392e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 0.662399674895865,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.3029550230900812e-05,
+      "loss": 0.1997,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6634156253174845,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2958842573996016e-05,
+      "loss": 0.1969,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6644315757391039,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2888260110711525e-05,
+      "loss": 0.1469,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6654475261607233,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.2817803574903212e-05,
+      "loss": 0.1524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6664634765823427,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2747473699117668e-05,
+      "loss": 0.159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6674794270039622,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.267727121458458e-05,
+      "loss": 0.1999,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6684953774255816,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.2607196851209137e-05,
+      "loss": 0.2216,
+      "step": 6580
+    },
+    {
+      "epoch": 0.669511327847201,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2537251337564412e-05,
+      "loss": 0.1607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6705272782688205,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.2467435400883839e-05,
+      "loss": 0.2187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6715432286904399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.239774976705359e-05,
+      "loss": 0.1753,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6725591791120593,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2328195160605092e-05,
+      "loss": 0.194,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6735751295336787,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.225877230470743e-05,
+      "loss": 0.1485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6745910799552982,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.218948192115988e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6756070303769176,
+      "grad_norm": 3.875,
+      "learning_rate": 1.21203247303844e-05,
+      "loss": 0.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.676622980798537,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.2051301451418073e-05,
+      "loss": 0.2377,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6776389312201565,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.198241280190574e-05,
+      "loss": 0.1508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6786548816417759,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.1913659498092431e-05,
+      "loss": 0.1537,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6796708320633953,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.184504225481601e-05,
+      "loss": 0.2339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6806867824850147,
+      "grad_norm": 5.625,
+      "learning_rate": 1.177656178549966e-05,
+      "loss": 0.2102,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6817027329066342,
+      "grad_norm": 2.5,
+      "learning_rate": 1.1708218802144536e-05,
+      "loss": 0.1435,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6827186833282536,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1640014015322323e-05,
+      "loss": 0.1823,
+      "step": 6720
+    },
+    {
+      "epoch": 0.683734633749873,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1571948134167862e-05,
+      "loss": 0.1154,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6847505841714925,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1504021866371761e-05,
+      "loss": 0.2105,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6857665345931119,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.143623591817304e-05,
+      "loss": 0.1317,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6867824850147313,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.1368590994351835e-05,
+      "loss": 0.1406,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6877984354363507,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.130108779822198e-05,
+      "loss": 0.1425,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6888143858579702,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.1233727031623783e-05,
+      "loss": 0.1623,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6898303362795896,
+      "grad_norm": 4.625,
+      "learning_rate": 1.1166509394916682e-05,
+      "loss": 0.1591,
+      "step": 6790
+    },
+    {
+      "epoch": 0.690846286701209,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1099435586971982e-05,
+      "loss": 0.1758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6918622371228285,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.1032506305165555e-05,
+      "loss": 0.1018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6928781875444479,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.0965722245370641e-05,
+      "loss": 0.1485,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6938941379660672,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0899084101950561e-05,
+      "loss": 0.1762,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6949100883876866,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.0832592567751555e-05,
+      "loss": 0.1402,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6959260388093061,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0766248334095505e-05,
+      "loss": 0.2278,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6969419892309255,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.0700052090772828e-05,
+      "loss": 0.1969,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6979579396525449,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.0634004526035249e-05,
+      "loss": 0.2073,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6989738900741643,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.0568106326588645e-05,
+      "loss": 0.1902,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6999898404957838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.0502358177585953e-05,
+      "loss": 0.2165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7010057909174032,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.0436760762619977e-05,
+      "loss": 0.1952,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7020217413390226,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.0371314763716347e-05,
+      "loss": 0.1422,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7030376917606421,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.0306020861326388e-05,
+      "loss": 0.0961,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7040536421822615,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.0240879734320068e-05,
+      "loss": 0.1542,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7050695926038809,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.0175892059978901e-05,
+      "loss": 0.1748,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7060855430255003,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.0111058513988958e-05,
+      "loss": 0.0819,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7071014934471198,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.0046379770433803e-05,
+      "loss": 0.1933,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7081174438687392,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.98185650178749e-06,
+      "loss": 0.1891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7091333942903586,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.917489378907591e-06,
+      "loss": 0.2102,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7101493447119781,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.853279071028212e-06,
+      "loss": 0.1714,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7111652951335975,
+      "grad_norm": 2.375,
+      "learning_rate": 9.78922624575303e-06,
+      "loss": 0.1299,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7121812455552169,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.72533156904833e-06,
+      "loss": 0.1914,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7131971959768363,
+      "grad_norm": 3.859375,
+      "learning_rate": 9.661595705236137e-06,
+      "loss": 0.2377,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7142131463984558,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.598019316987244e-06,
+      "loss": 0.1851,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7152290968200752,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.53460306531439e-06,
+      "loss": 0.2661,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7162450472416946,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.471347609565311e-06,
+      "loss": 0.1669,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7172609976633141,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.408253607415957e-06,
+      "loss": 0.2487,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7182769480849335,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.345321714863614e-06,
+      "loss": 0.186,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7192928985065529,
+      "grad_norm": 6.0625,
+      "learning_rate": 9.282552586220075e-06,
+      "loss": 0.2249,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7203088489281723,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.219946874104885e-06,
+      "loss": 0.1255,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7213247993497918,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.157505229438481e-06,
+      "loss": 0.1999,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7223407497714112,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.095228301435518e-06,
+      "loss": 0.199,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7233567001930306,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.03311673759802e-06,
+      "loss": 0.2182,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7243726506146501,
+      "grad_norm": 6.46875,
+      "learning_rate": 8.971171183708733e-06,
+      "loss": 0.1573,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7253886010362695,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.909392283824353e-06,
+      "loss": 0.2044,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7264045514578888,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.847780680268872e-06,
+      "loss": 0.11,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7274205018795082,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.786337013626853e-06,
+      "loss": 0.1897,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7284364523011277,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.725061922736799e-06,
+      "loss": 0.153,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7294524027227471,
+      "grad_norm": 1.609375,
+      "learning_rate": 8.663956044684532e-06,
+      "loss": 0.1746,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7304683531443665,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.603020014796507e-06,
+      "loss": 0.2284,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7314843035659859,
+      "grad_norm": 1.515625,
+      "learning_rate": 8.542254466633273e-06,
+      "loss": 0.1186,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7325002539876054,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.481660031982844e-06,
+      "loss": 0.1971,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7335162044092248,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.421237340854157e-06,
+      "loss": 0.196,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7345321548308442,
+      "grad_norm": 0.65234375,
+      "learning_rate": 8.360987021470479e-06,
+      "loss": 0.1724,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7355481052524637,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.300909700262929e-06,
+      "loss": 0.175,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7365640556740831,
+      "grad_norm": 3.109375,
+      "learning_rate": 8.241006001863924e-06,
+      "loss": 0.2276,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7375800060957025,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.181276549100714e-06,
+      "loss": 0.2029,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7385959565173219,
+      "grad_norm": 4.03125,
+      "learning_rate": 8.12172196298887e-06,
+      "loss": 0.175,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7396119069389414,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.062342862725878e-06,
+      "loss": 0.1662,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7406278573605608,
+      "grad_norm": 3.375,
+      "learning_rate": 8.003139865684662e-06,
+      "loss": 0.1616,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7416438077821802,
+      "grad_norm": 2.5625,
+      "learning_rate": 7.944113587407157e-06,
+      "loss": 0.2448,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7426597582037997,
+      "grad_norm": 4.125,
+      "learning_rate": 7.885264641597961e-06,
+      "loss": 0.1618,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7436757086254191,
+      "grad_norm": 3.5,
+      "learning_rate": 7.826593640117889e-06,
+      "loss": 0.1134,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7446916590470385,
+      "grad_norm": 2.6875,
+      "learning_rate": 7.76810119297767e-06,
+      "loss": 0.1795,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7457076094686579,
+      "grad_norm": 4.34375,
+      "learning_rate": 7.709787908331556e-06,
+      "loss": 0.2736,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7467235598902774,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.651654392471038e-06,
+      "loss": 0.139,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7477395103118968,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.593701249818521e-06,
+      "loss": 0.2023,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7487554607335162,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.535929082921048e-06,
+      "loss": 0.1702,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7497714111551357,
+      "grad_norm": 1.96875,
+      "learning_rate": 7.47833849244402e-06,
+      "loss": 0.1835,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7507873615767551,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.420930077164959e-06,
+      "loss": 0.1713,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7518033119983745,
+      "grad_norm": 4.46875,
+      "learning_rate": 7.363704433967311e-06,
+      "loss": 0.1906,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7528192624199939,
+      "grad_norm": 1.75,
+      "learning_rate": 7.306662157834185e-06,
+      "loss": 0.1421,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7538352128416134,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.2498038418422145e-06,
+      "loss": 0.1793,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7548511632632328,
+      "grad_norm": 2.578125,
+      "learning_rate": 7.193130077155374e-06,
+      "loss": 0.1603,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7558671136848522,
+      "grad_norm": 4.3125,
+      "learning_rate": 7.13664145301883e-06,
+      "loss": 0.2169,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7568830641064717,
+      "grad_norm": 3.078125,
+      "learning_rate": 7.0803385567528025e-06,
+      "loss": 0.1685,
+      "step": 7450
+    },
+    {
+      "epoch": 0.757899014528091,
+      "grad_norm": 3.5625,
+      "learning_rate": 7.024221973746495e-06,
+      "loss": 0.2282,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7589149649497104,
+      "grad_norm": 2.265625,
+      "learning_rate": 6.968292287451961e-06,
+      "loss": 0.1786,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7599309153713298,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.912550079378091e-06,
+      "loss": 0.1811,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7609468657929493,
+      "grad_norm": 2.328125,
+      "learning_rate": 6.856995929084506e-06,
+      "loss": 0.1747,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7619628162145687,
+      "grad_norm": 5.21875,
+      "learning_rate": 6.801630414175589e-06,
+      "loss": 0.2028,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7629787666361881,
+      "grad_norm": 3.78125,
+      "learning_rate": 6.746454110294451e-06,
+      "loss": 0.2255,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7639947170578075,
+      "grad_norm": 1.625,
+      "learning_rate": 6.691467591116931e-06,
+      "loss": 0.1604,
+      "step": 7520
+    },
+    {
+      "epoch": 0.765010667479427,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.6366714283456755e-06,
+      "loss": 0.2559,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7660266179010464,
+      "grad_norm": 4.59375,
+      "learning_rate": 6.582066191704142e-06,
+      "loss": 0.2034,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7670425683226658,
+      "grad_norm": 1.578125,
+      "learning_rate": 6.527652448930724e-06,
+      "loss": 0.148,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7680585187442853,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.4734307657728e-06,
+      "loss": 0.1811,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7690744691659047,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.419401705980924e-06,
+      "loss": 0.1407,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7700904195875241,
+      "grad_norm": 2.25,
+      "learning_rate": 6.365565831302869e-06,
+      "loss": 0.1893,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7711063700091435,
+      "grad_norm": 1.625,
+      "learning_rate": 6.311923701477854e-06,
+      "loss": 0.1835,
+      "step": 7590
+    },
+    {
+      "epoch": 0.772122320430763,
+      "grad_norm": 2.375,
+      "learning_rate": 6.258475874230713e-06,
+      "loss": 0.1579,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7731382708523824,
+      "grad_norm": 4.5,
+      "learning_rate": 6.205222905266067e-06,
+      "loss": 0.1794,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7741542212740018,
+      "grad_norm": 4.25,
+      "learning_rate": 6.152165348262598e-06,
+      "loss": 0.1477,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7751701716956213,
+      "grad_norm": 1.9765625,
+      "learning_rate": 6.0993037548672246e-06,
+      "loss": 0.2396,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7761861221172407,
+      "grad_norm": 2.671875,
+      "learning_rate": 6.046638674689454e-06,
+      "loss": 0.1717,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7772020725388601,
+      "grad_norm": 3.671875,
+      "learning_rate": 5.994170655295567e-06,
+      "loss": 0.2646,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7782180229604795,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.9419002422030106e-06,
+      "loss": 0.1553,
+      "step": 7660
+    },
+    {
+      "epoch": 0.779233973382099,
+      "grad_norm": 3.734375,
+      "learning_rate": 5.889827978874665e-06,
+      "loss": 0.1854,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7802499238037184,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.837954406713245e-06,
+      "loss": 0.1857,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7812658742253378,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.786280065055619e-06,
+      "loss": 0.1797,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7822818246469573,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.734805491167244e-06,
+      "loss": 0.1488,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7832977750685767,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.683531220236576e-06,
+      "loss": 0.1688,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.632457785369455e-06,
+      "loss": 0.1503,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7853296759118155,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.581585717583637e-06,
+      "loss": 0.1658,
+      "step": 7730
+    },
+    {
+      "epoch": 0.786345626333435,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.530915545803209e-06,
+      "loss": 0.2112,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7873615767550544,
+      "grad_norm": 4.1875,
+      "learning_rate": 5.480447796853141e-06,
+      "loss": 0.165,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7883775271766738,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.430182995453756e-06,
+      "loss": 0.1499,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7893934775982933,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.380121664215329e-06,
+      "loss": 0.1559,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7904094280199127,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.330264323632611e-06,
+      "loss": 0.2098,
+      "step": 7780
+    },
+    {
+      "epoch": 0.791425378441532,
+      "grad_norm": 4.65625,
+      "learning_rate": 5.280611492079449e-06,
+      "loss": 0.1776,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7924413288631514,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.231163685803361e-06,
+      "loss": 0.1497,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7934572792847709,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.181921418920191e-06,
+      "loss": 0.12,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7944732297063903,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.13288520340878e-06,
+      "loss": 0.1981,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7954891801280097,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.084055549105596e-06,
+      "loss": 0.1389,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7965051305496291,
+      "grad_norm": 2.796875,
+      "learning_rate": 5.035432963699479e-06,
+      "loss": 0.2293,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7975210809712486,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.98701795272635e-06,
+      "loss": 0.1618,
+      "step": 7850
+    },
+    {
+      "epoch": 0.798537031392868,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.938811019563938e-06,
+      "loss": 0.1755,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7995529818144874,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.8908126654265475e-06,
+      "loss": 0.1565,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8005689322361069,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.843023389359885e-06,
+      "loss": 0.2176,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8015848826577263,
+      "grad_norm": 2.625,
+      "learning_rate": 4.79544368823581e-06,
+      "loss": 0.2013,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8026008330793457,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.748074056747234e-06,
+      "loss": 0.1246,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8036167835009651,
+      "grad_norm": 3.5,
+      "learning_rate": 4.700914987402919e-06,
+      "loss": 0.1638,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8046327339225846,
+      "grad_norm": 3.4375,
+      "learning_rate": 4.6539669705223916e-06,
+      "loss": 0.2213,
+      "step": 7920
+    },
+    {
+      "epoch": 0.805648684344204,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.607230494230849e-06,
+      "loss": 0.1822,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8066646347658234,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.560706044454047e-06,
+      "loss": 0.1763,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8076805851874429,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.514394104913291e-06,
+      "loss": 0.234,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8086965356090623,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.468295157120372e-06,
+      "loss": 0.1939,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8097124860306817,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.422409680372594e-06,
+      "loss": 0.174,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8107284364523011,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.3767381517477505e-06,
+      "loss": 0.2375,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8117443868739206,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.331281046099203e-06,
+      "loss": 0.2076,
+      "step": 7990
+    },
+    {
+      "epoch": 0.81276033729554,
+      "grad_norm": 6.0625,
+      "learning_rate": 4.286038836050929e-06,
+      "loss": 0.2504,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8137762877171594,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.241011991992586e-06,
+      "loss": 0.2102,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8147922381387789,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.1962009820746635e-06,
+      "loss": 0.1846,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8158081885603983,
+      "grad_norm": 1.875,
+      "learning_rate": 4.15160627220357e-06,
+      "loss": 0.1741,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8168241389820177,
+      "grad_norm": 5.5625,
+      "learning_rate": 4.107228326036838e-06,
+      "loss": 0.2078,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8178400894036371,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.063067604978252e-06,
+      "loss": 0.212,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8188560398252566,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.019124568173094e-06,
+      "loss": 0.1831,
+      "step": 8060
+    },
+    {
+      "epoch": 0.819871990246876,
+      "grad_norm": 6.625,
+      "learning_rate": 3.975399672503341e-06,
+      "loss": 0.2196,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8208879406684954,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.931893372582943e-06,
+      "loss": 0.2002,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8219038910901149,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.888606120753047e-06,
+      "loss": 0.2138,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8229198415117343,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.845538367077362e-06,
+      "loss": 0.2593,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8239357919333536,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.8026905593374213e-06,
+      "loss": 0.2062,
+      "step": 8110
+    },
+    {
+      "epoch": 0.824951742354973,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.760063143027945e-06,
+      "loss": 0.1343,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8259676927765925,
+      "grad_norm": 1.984375,
+      "learning_rate": 3.7176565613522313e-06,
+      "loss": 0.2494,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8269836431982119,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.675471255217516e-06,
+      "loss": 0.1502,
+      "step": 8140
+    },
+    {
+      "epoch": 0.8279995936198313,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.6335076632304175e-06,
+      "loss": 0.1256,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8290155440414507,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.5917662216923332e-06,
+      "loss": 0.1709,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8300314944630702,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.550247364594958e-06,
+      "loss": 0.1881,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8310474448846896,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.508951523615725e-06,
+      "loss": 0.1998,
+      "step": 8180
+    },
+    {
+      "epoch": 0.832063395306309,
+      "grad_norm": 2.40625,
+      "learning_rate": 3.467879128113352e-06,
+      "loss": 0.2429,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8330793457279285,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.427030605123352e-06,
+      "loss": 0.1942,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8340952961495479,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.3864063793536043e-06,
+      "loss": 0.1898,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8351112465711673,
+      "grad_norm": 5.375,
+      "learning_rate": 3.3460068731799577e-06,
+      "loss": 0.1919,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8361271969927867,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.3058325066417818e-06,
+      "loss": 0.1516,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8371431474144062,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.26588369743768e-06,
+      "loss": 0.1068,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8381590978360256,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.2261608609210653e-06,
+      "loss": 0.1203,
+      "step": 8250
+    },
+    {
+      "epoch": 0.839175048257645,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.186664410095913e-06,
+      "loss": 0.2172,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8401909986792645,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.1473947556124093e-06,
+      "loss": 0.1249,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8412069491008839,
+      "grad_norm": 2.484375,
+      "learning_rate": 3.1083523057627213e-06,
+      "loss": 0.1744,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8422228995225033,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.0695374664767353e-06,
+      "loss": 0.1772,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8432388499441227,
+      "grad_norm": 0.59375,
+      "learning_rate": 3.0309506413178397e-06,
+      "loss": 0.2302,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8442548003657422,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.9925922314787136e-06,
+      "loss": 0.1635,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8452707507873616,
+      "grad_norm": 2.34375,
+      "learning_rate": 2.954462635777194e-06,
+      "loss": 0.1573,
+      "step": 8320
+    },
+    {
+      "epoch": 0.846286701208981,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.916562250652083e-06,
+      "loss": 0.1608,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8473026516306005,
+      "grad_norm": 4.125,
+      "learning_rate": 2.878891470159048e-06,
+      "loss": 0.184,
+      "step": 8340
+    },
+    {
+      "epoch": 0.8483186020522199,
+      "grad_norm": 2.515625,
+      "learning_rate": 2.8414506859665514e-06,
+      "loss": 0.2141,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8493345524738393,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8042402873517197e-06,
+      "loss": 0.1729,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8503505028954587,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.76726066119635e-06,
+      "loss": 0.2252,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8513664533170782,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.730512191982845e-06,
+      "loss": 0.1644,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8523824037386976,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.693995261790261e-06,
+      "loss": 0.1822,
+      "step": 8390
+    },
+    {
+      "epoch": 0.853398354160317,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.657710250290285e-06,
+      "loss": 0.2068,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8544143045819365,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.621657534743327e-06,
+      "loss": 0.1224,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8554302550035559,
+      "grad_norm": 3.421875,
+      "learning_rate": 2.5858374899945804e-06,
+      "loss": 0.179,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8564462054251752,
+      "grad_norm": 3.484375,
+      "learning_rate": 2.550250488470135e-06,
+      "loss": 0.1873,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8574621558467946,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.5148969001730806e-06,
+      "loss": 0.1799,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8584781062684141,
+      "grad_norm": 1.375,
+      "learning_rate": 2.4797770926796858e-06,
+      "loss": 0.176,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8594940566900335,
+      "grad_norm": 1.8984375,
+      "learning_rate": 2.444891431135571e-06,
+      "loss": 0.1664,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8605100071116529,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.4102402782518936e-06,
+      "loss": 0.1512,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8615259575332723,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.3758239943016096e-06,
+      "loss": 0.1629,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8625419079548918,
+      "grad_norm": 5.3125,
+      "learning_rate": 2.3416429371157013e-06,
+      "loss": 0.2099,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8635578583765112,
+      "grad_norm": 5.9375,
+      "learning_rate": 2.307697462079464e-06,
+      "loss": 0.2221,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8645738087981306,
+      "grad_norm": 5.4375,
+      "learning_rate": 2.273987922128809e-06,
+      "loss": 0.2191,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8655897592197501,
+      "grad_norm": 2.171875,
+      "learning_rate": 2.240514667746607e-06,
+      "loss": 0.1843,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8666057096413695,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.2072780469590245e-06,
+      "loss": 0.2494,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8676216600629889,
+      "grad_norm": 2.25,
+      "learning_rate": 2.1742784053319116e-06,
+      "loss": 0.1712,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8686376104846083,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.141516085967224e-06,
+      "loss": 0.1169,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8696535609062278,
+      "grad_norm": 4.25,
+      "learning_rate": 2.1089914294994434e-06,
+      "loss": 0.1374,
+      "step": 8560
+    },
+    {
+      "epoch": 0.8706695113278472,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.0767047740920336e-06,
+      "loss": 0.2162,
+      "step": 8570
+    },
+    {
+      "epoch": 0.8716854617494666,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.0446564554339187e-06,
+      "loss": 0.1593,
+      "step": 8580
+    },
+    {
+      "epoch": 0.8727014121710861,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.0128468067360185e-06,
+      "loss": 0.1857,
+      "step": 8590
+    },
+    {
+      "epoch": 0.8737173625927055,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.981276158727749e-06,
+      "loss": 0.1989,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8747333130143249,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.949944839653625e-06,
+      "loss": 0.2077,
+      "step": 8610
+    },
+    {
+      "epoch": 0.8757492634359443,
+      "grad_norm": 2.625,
+      "learning_rate": 1.918853175269797e-06,
+      "loss": 0.2003,
+      "step": 8620
+    },
+    {
+      "epoch": 0.8767652138575638,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.8880014888407127e-06,
+      "loss": 0.2486,
+      "step": 8630
+    },
+    {
+      "epoch": 0.8777811642791832,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.8573901011357336e-06,
+      "loss": 0.1896,
+      "step": 8640
+    },
+    {
+      "epoch": 0.8787971147008026,
+      "grad_norm": 5.0625,
+      "learning_rate": 1.8270193304257887e-06,
+      "loss": 0.1727,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8798130651224221,
+      "grad_norm": 1.75,
+      "learning_rate": 1.7968894924800916e-06,
+      "loss": 0.1687,
+      "step": 8660
+    },
+    {
+      "epoch": 0.8808290155440415,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.7670009005628291e-06,
+      "loss": 0.166,
+      "step": 8670
+    },
+    {
+      "epoch": 0.8818449659656609,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.737353865429936e-06,
+      "loss": 0.1471,
+      "step": 8680
+    },
+    {
+      "epoch": 0.8828609163872803,
+      "grad_norm": 0.546875,
+      "learning_rate": 1.7079486953258283e-06,
+      "loss": 0.1075,
+      "step": 8690
+    },
+    {
+      "epoch": 0.8838768668088998,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.6787856959802367e-06,
+      "loss": 0.2113,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8848928172305192,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.6498651706049945e-06,
+      "loss": 0.1412,
+      "step": 8710
+    },
+    {
+      "epoch": 0.8859087676521386,
+      "grad_norm": 3.796875,
+      "learning_rate": 1.6211874198909072e-06,
+      "loss": 0.1701,
+      "step": 8720
+    },
+    {
+      "epoch": 0.8869247180737581,
+      "grad_norm": 3.734375,
+      "learning_rate": 1.592752742004605e-06,
+      "loss": 0.1348,
+      "step": 8730
+    },
+    {
+      "epoch": 0.8879406684953774,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.5645614325854735e-06,
+      "loss": 0.1931,
+      "step": 8740
+    },
+    {
+      "epoch": 0.8889566189169968,
+      "grad_norm": 3.4375,
+      "learning_rate": 1.5366137847425466e-06,
+      "loss": 0.1705,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8899725693386162,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.5089100890514769e-06,
+      "loss": 0.1889,
+      "step": 8760
+    },
+    {
+      "epoch": 0.8909885197602357,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.4814506335515176e-06,
+      "loss": 0.1837,
+      "step": 8770
+    },
+    {
+      "epoch": 0.8920044701818551,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.4542357037425207e-06,
+      "loss": 0.1728,
+      "step": 8780
+    },
+    {
+      "epoch": 0.8930204206034745,
+      "grad_norm": 1.625,
+      "learning_rate": 1.4272655825819713e-06,
+      "loss": 0.1562,
+      "step": 8790
+    },
+    {
+      "epoch": 0.8940363710250939,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.4005405504820351e-06,
+      "loss": 0.1681,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8950523214467134,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.3740608853066634e-06,
+      "loss": 0.1449,
+      "step": 8810
+    },
+    {
+      "epoch": 0.8960682718683328,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.347826862368684e-06,
+      "loss": 0.2418,
+      "step": 8820
+    },
+    {
+      "epoch": 0.8970842222899522,
+      "grad_norm": 0.55859375,
+      "learning_rate": 1.3218387544269545e-06,
+      "loss": 0.2473,
+      "step": 8830
+    },
+    {
+      "epoch": 0.8981001727115717,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.2960968316835132e-06,
+      "loss": 0.194,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8991161231331911,
+      "grad_norm": 3.921875,
+      "learning_rate": 1.2706013617807822e-06,
+      "loss": 0.2109,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9001320735548105,
+      "grad_norm": 5.03125,
+      "learning_rate": 1.2453526097987778e-06,
+      "loss": 0.151,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9011480239764299,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.2203508382523431e-06,
+      "loss": 0.1811,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9021639743980494,
+      "grad_norm": 3.828125,
+      "learning_rate": 1.1955963070884534e-06,
+      "loss": 0.2004,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9031799248196688,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.171089273683465e-06,
+      "loss": 0.1395,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9041958752412882,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.1468299928404868e-06,
+      "loss": 0.1915,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9052118256629077,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.1228187167866943e-06,
+      "loss": 0.1281,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9062277760845271,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.099055695170728e-06,
+      "loss": 0.1627,
+      "step": 8920
+    },
+    {
+      "epoch": 0.9072437265061465,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.0755411750600962e-06,
+      "loss": 0.1768,
+      "step": 8930
+    },
+    {
+      "epoch": 0.9082596769277659,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.052275400938596e-06,
+      "loss": 0.1544,
+      "step": 8940
+    },
+    {
+      "epoch": 0.9092756273493854,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.0292586147037764e-06,
+      "loss": 0.2498,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9102915777710048,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.0064910556644214e-06,
+      "loss": 0.1918,
+      "step": 8960
+    },
+    {
+      "epoch": 0.9113075281926242,
+      "grad_norm": 4.0,
+      "learning_rate": 9.839729605380766e-07,
+      "loss": 0.2388,
+      "step": 8970
+    },
+    {
+      "epoch": 0.9123234786142437,
+      "grad_norm": 3.765625,
+      "learning_rate": 9.61704563448565e-07,
+      "loss": 0.1944,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9133394290358631,
+      "grad_norm": 2.90625,
+      "learning_rate": 9.396860959235671e-07,
+      "loss": 0.1667,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9143553794574825,
+      "grad_norm": 2.4375,
+      "learning_rate": 9.179177868922085e-07,
+      "loss": 0.2143,
+      "step": 9000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-9000/training_args.bin b/checkpoints/checkpoint-9000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-9000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/checkpoints/checkpoint-9500/adapter_config.json b/checkpoints/checkpoint-9500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..2f24c7f88c4378c16a55c7a5a7e4079b1e95ac70
--- /dev/null
+++ b/checkpoints/checkpoint-9500/adapter_config.json
@@ -0,0 +1,29 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": "gaussian",
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0.1,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoints/checkpoint-9500/adapter_model.safetensors b/checkpoints/checkpoint-9500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4873d7ac0dd93d300c5305928706f0491b7f09e2
--- /dev/null
+++ b/checkpoints/checkpoint-9500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2b633ee47b5f3a143fc9c71b1dd0cfeb40cc097a762104327a42af91522c1dc
+size 5919456
diff --git a/checkpoints/checkpoint-9500/optimizer.pt b/checkpoints/checkpoint-9500/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d8453ac00ba64242e454f9493f2b7f0058728a8c
--- /dev/null
+++ b/checkpoints/checkpoint-9500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ada548d4424bd56d11910556a417182e99459e8c5770ee2fd7a75b4cf595f8b5
+size 11930938
diff --git a/checkpoints/checkpoint-9500/rng_state_0.pth b/checkpoints/checkpoint-9500/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..a76a2679c5ec6f028f79881ea7b86547e50b925c
--- /dev/null
+++ b/checkpoints/checkpoint-9500/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdd3431862bb3caf99bbe632d5500d35b3a4001ba5f79dfb311597988ce7bc07
+size 15024
diff --git a/checkpoints/checkpoint-9500/rng_state_1.pth b/checkpoints/checkpoint-9500/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d638e87f5f6cf02dbf4bd50290706a308924bd9f
--- /dev/null
+++ b/checkpoints/checkpoint-9500/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a49b137689bb9baeb88d62569411ec42b18c43387181a8e482c1188959e80ba8
+size 15024
diff --git a/checkpoints/checkpoint-9500/rng_state_2.pth b/checkpoints/checkpoint-9500/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..44200b445027eedae447f8c9abec1fa8bfd8922f
--- /dev/null
+++ b/checkpoints/checkpoint-9500/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f64f0a9a34ed397df06f2a99347433818ab578da206dd635ee4d819320c29732
+size 15024
diff --git a/checkpoints/checkpoint-9500/rng_state_3.pth b/checkpoints/checkpoint-9500/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2dd77ff9ed7fd897ba735ad0f9af28473a4e29c2
--- /dev/null
+++ b/checkpoints/checkpoint-9500/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f2446089864551b4466c0d305e103a1a414455ac1dd2cffffdee7d5b89d904a
+size 15024
diff --git a/checkpoints/checkpoint-9500/scheduler.pt b/checkpoints/checkpoint-9500/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..294d12a69b8fec0e3bf520b584bc6bb862f59d5d
--- /dev/null
+++ b/checkpoints/checkpoint-9500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c40a1ffb822b62d125ac636a29ddd80b085b0fed5f9760c5a078bde337791b1a
+size 1064
diff --git a/checkpoints/checkpoint-9500/trainer_state.json b/checkpoints/checkpoint-9500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..92150ac85879e6f7310f59e251c296183fd37122
--- /dev/null
+++ b/checkpoints/checkpoint-9500/trainer_state.json
@@ -0,0 +1,6671 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9651529005384537,
+  "eval_steps": 500,
+  "global_step": 9500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001015950421619425,
+      "grad_norm": 17.625,
+      "learning_rate": 5e-06,
+      "loss": 3.4264,
+      "step": 10
+    },
+    {
+      "epoch": 0.00203190084323885,
+      "grad_norm": 12.5625,
+      "learning_rate": 1e-05,
+      "loss": 3.432,
+      "step": 20
+    },
+    {
+      "epoch": 0.003047851264858275,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5e-05,
+      "loss": 3.23,
+      "step": 30
+    },
+    {
+      "epoch": 0.0040638016864777,
+      "grad_norm": 12.4375,
+      "learning_rate": 2e-05,
+      "loss": 2.9762,
+      "step": 40
+    },
+    {
+      "epoch": 0.005079752108097125,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.5e-05,
+      "loss": 2.6173,
+      "step": 50
+    },
+    {
+      "epoch": 0.00609570252971655,
+      "grad_norm": 10.1875,
+      "learning_rate": 3e-05,
+      "loss": 2.2004,
+      "step": 60
+    },
+    {
+      "epoch": 0.007111652951335975,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4176,
+      "step": 70
+    },
+    {
+      "epoch": 0.0081276033729554,
+      "grad_norm": 4.375,
+      "learning_rate": 4e-05,
+      "loss": 1.0122,
+      "step": 80
+    },
+    {
+      "epoch": 0.009143553794574825,
+      "grad_norm": 6.5625,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9116,
+      "step": 90
+    },
+    {
+      "epoch": 0.01015950421619425,
+      "grad_norm": 5.28125,
+      "learning_rate": 5e-05,
+      "loss": 0.6832,
+      "step": 100
+    },
+    {
+      "epoch": 0.011175454637813675,
+      "grad_norm": 5.5,
+      "learning_rate": 4.9999870035728426e-05,
+      "loss": 0.7355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0121914050594331,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.9999480144264944e-05,
+      "loss": 0.6673,
+      "step": 120
+    },
+    {
+      "epoch": 0.013207355481052525,
+      "grad_norm": 4.5,
+      "learning_rate": 4.9998830329663314e-05,
+      "loss": 0.6792,
+      "step": 130
+    },
+    {
+      "epoch": 0.01422330590267195,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.9997920598679756e-05,
+      "loss": 0.6207,
+      "step": 140
+    },
+    {
+      "epoch": 0.015239256324291375,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.999675096077286e-05,
+      "loss": 0.483,
+      "step": 150
+    },
+    {
+      "epoch": 0.0162552067459108,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.999532142810354e-05,
+      "loss": 0.5319,
+      "step": 160
+    },
+    {
+      "epoch": 0.017271157167530225,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.999363201553483e-05,
+      "loss": 0.6052,
+      "step": 170
+    },
+    {
+      "epoch": 0.01828710758914965,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.9991682740631794e-05,
+      "loss": 0.4258,
+      "step": 180
+    },
+    {
+      "epoch": 0.019303058010769075,
+      "grad_norm": 3.859375,
+      "learning_rate": 4.998947362366133e-05,
+      "loss": 0.4309,
+      "step": 190
+    },
+    {
+      "epoch": 0.0203190084323885,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.998700468759193e-05,
+      "loss": 0.3957,
+      "step": 200
+    },
+    {
+      "epoch": 0.021334958854007924,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9984275958093475e-05,
+      "loss": 0.4777,
+      "step": 210
+    },
+    {
+      "epoch": 0.02235090927562735,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.998128746353695e-05,
+      "loss": 0.3549,
+      "step": 220
+    },
+    {
+      "epoch": 0.023366859697246774,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.997803923499417e-05,
+      "loss": 0.4447,
+      "step": 230
+    },
+    {
+      "epoch": 0.0243828101188662,
+      "grad_norm": 6.375,
+      "learning_rate": 4.99745313062374e-05,
+      "loss": 0.3808,
+      "step": 240
+    },
+    {
+      "epoch": 0.025398760540485624,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.99707637137391e-05,
+      "loss": 0.3827,
+      "step": 250
+    },
+    {
+      "epoch": 0.02641471096210505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.996673649667145e-05,
+      "loss": 0.3694,
+      "step": 260
+    },
+    {
+      "epoch": 0.027430661383724474,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.9962449696906e-05,
+      "loss": 0.3586,
+      "step": 270
+    },
+    {
+      "epoch": 0.0284466118053439,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9957903359013214e-05,
+      "loss": 0.3832,
+      "step": 280
+    },
+    {
+      "epoch": 0.029462562226963324,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.995309753026201e-05,
+      "loss": 0.328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03047851264858275,
+      "grad_norm": 4.5,
+      "learning_rate": 4.994803226061927e-05,
+      "loss": 0.3667,
+      "step": 300
+    },
+    {
+      "epoch": 0.03149446307020217,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.994270760274933e-05,
+      "loss": 0.3811,
+      "step": 310
+    },
+    {
+      "epoch": 0.0325104134918216,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.99371236120134e-05,
+      "loss": 0.3065,
+      "step": 320
+    },
+    {
+      "epoch": 0.03352636391344102,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.993128034646902e-05,
+      "loss": 0.4177,
+      "step": 330
+    },
+    {
+      "epoch": 0.03454231433506045,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.992517786686947e-05,
+      "loss": 0.33,
+      "step": 340
+    },
+    {
+      "epoch": 0.03555826475667987,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.9918816236663077e-05,
+      "loss": 0.3287,
+      "step": 350
+    },
+    {
+      "epoch": 0.0365742151782993,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.991219552199262e-05,
+      "loss": 0.2934,
+      "step": 360
+    },
+    {
+      "epoch": 0.03759016559991872,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.99053157916946e-05,
+      "loss": 0.3176,
+      "step": 370
+    },
+    {
+      "epoch": 0.03860611602153815,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.989817711729856e-05,
+      "loss": 0.3318,
+      "step": 380
+    },
+    {
+      "epoch": 0.03962206644315757,
+      "grad_norm": 2.375,
+      "learning_rate": 4.98907795730263e-05,
+      "loss": 0.3234,
+      "step": 390
+    },
+    {
+      "epoch": 0.040638016864777,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.988312323579114e-05,
+      "loss": 0.267,
+      "step": 400
+    },
+    {
+      "epoch": 0.04165396728639642,
+      "grad_norm": 3.75,
+      "learning_rate": 4.98752081851971e-05,
+      "loss": 0.3081,
+      "step": 410
+    },
+    {
+      "epoch": 0.04266991770801585,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.986703450353809e-05,
+      "loss": 0.2917,
+      "step": 420
+    },
+    {
+      "epoch": 0.04368586812963527,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.985860227579703e-05,
+      "loss": 0.2805,
+      "step": 430
+    },
+    {
+      "epoch": 0.0447018185512547,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.984991158964499e-05,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.04571776897287412,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.9840962535440265e-05,
+      "loss": 0.335,
+      "step": 450
+    },
+    {
+      "epoch": 0.04673371939449355,
+      "grad_norm": 3.25,
+      "learning_rate": 4.983175520622744e-05,
+      "loss": 0.2544,
+      "step": 460
+    },
+    {
+      "epoch": 0.04774966981611297,
+      "grad_norm": 2.25,
+      "learning_rate": 4.982228969773642e-05,
+      "loss": 0.3449,
+      "step": 470
+    },
+    {
+      "epoch": 0.0487656202377324,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.9812566108381435e-05,
+      "loss": 0.2964,
+      "step": 480
+    },
+    {
+      "epoch": 0.04978157065935182,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9802584539260035e-05,
+      "loss": 0.2799,
+      "step": 490
+    },
+    {
+      "epoch": 0.05079752108097125,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.979234509415199e-05,
+      "loss": 0.3231,
+      "step": 500
+    },
+    {
+      "epoch": 0.05181347150259067,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.978184787951828e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 0.0528294219242101,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.977109300449992e-05,
+      "loss": 0.2705,
+      "step": 520
+    },
+    {
+      "epoch": 0.05384537234582952,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9760080580916876e-05,
+      "loss": 0.2998,
+      "step": 530
+    },
+    {
+      "epoch": 0.05486132276744895,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.974881072326688e-05,
+      "loss": 0.2595,
+      "step": 540
+    },
+    {
+      "epoch": 0.05587727318906837,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9737283548724236e-05,
+      "loss": 0.2803,
+      "step": 550
+    },
+    {
+      "epoch": 0.0568932236106878,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.97254991771386e-05,
+      "loss": 0.3511,
+      "step": 560
+    },
+    {
+      "epoch": 0.05790917403230722,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.971345773103377e-05,
+      "loss": 0.312,
+      "step": 570
+    },
+    {
+      "epoch": 0.05892512445392665,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9701159335606365e-05,
+      "loss": 0.2482,
+      "step": 580
+    },
+    {
+      "epoch": 0.05994107487554607,
+      "grad_norm": 5.5,
+      "learning_rate": 4.968860411872454e-05,
+      "loss": 0.2537,
+      "step": 590
+    },
+    {
+      "epoch": 0.0609570252971655,
+      "grad_norm": 3.546875,
+      "learning_rate": 4.967579221092666e-05,
+      "loss": 0.3125,
+      "step": 600
+    },
+    {
+      "epoch": 0.06197297571878492,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.966272374541996e-05,
+      "loss": 0.2354,
+      "step": 610
+    },
+    {
+      "epoch": 0.06298892614040434,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.964939885807912e-05,
+      "loss": 0.3213,
+      "step": 620
+    },
+    {
+      "epoch": 0.06400487656202378,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.9635817687444876e-05,
+      "loss": 0.3003,
+      "step": 630
+    },
+    {
+      "epoch": 0.0650208269836432,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.962198037472259e-05,
+      "loss": 0.2996,
+      "step": 640
+    },
+    {
+      "epoch": 0.06603677740526262,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.9607887063780776e-05,
+      "loss": 0.2257,
+      "step": 650
+    },
+    {
+      "epoch": 0.06705272782688204,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9593537901149564e-05,
+      "loss": 0.223,
+      "step": 660
+    },
+    {
+      "epoch": 0.06806867824850148,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.957893303601924e-05,
+      "loss": 0.3407,
+      "step": 670
+    },
+    {
+      "epoch": 0.0690846286701209,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.956407262023866e-05,
+      "loss": 0.2589,
+      "step": 680
+    },
+    {
+      "epoch": 0.07010057909174032,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.954895680831367e-05,
+      "loss": 0.2949,
+      "step": 690
+    },
+    {
+      "epoch": 0.07111652951335974,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.9533585757405506e-05,
+      "loss": 0.2995,
+      "step": 700
+    },
+    {
+      "epoch": 0.07213247993497918,
+      "grad_norm": 4.625,
+      "learning_rate": 4.951795962732917e-05,
+      "loss": 0.2894,
+      "step": 710
+    },
+    {
+      "epoch": 0.0731484303565986,
+      "grad_norm": 3.0,
+      "learning_rate": 4.9502078580551755e-05,
+      "loss": 0.3082,
+      "step": 720
+    },
+    {
+      "epoch": 0.07416438077821802,
+      "grad_norm": 3.65625,
+      "learning_rate": 4.9485942782190734e-05,
+      "loss": 0.2308,
+      "step": 730
+    },
+    {
+      "epoch": 0.07518033119983744,
+      "grad_norm": 4.78125,
+      "learning_rate": 4.9469552400012306e-05,
+      "loss": 0.2272,
+      "step": 740
+    },
+    {
+      "epoch": 0.07619628162145688,
+      "grad_norm": 4.25,
+      "learning_rate": 4.94529076044296e-05,
+      "loss": 0.2701,
+      "step": 750
+    },
+    {
+      "epoch": 0.0772122320430763,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.94360085685009e-05,
+      "loss": 0.2686,
+      "step": 760
+    },
+    {
+      "epoch": 0.07822818246469572,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.9418855467927894e-05,
+      "loss": 0.2051,
+      "step": 770
+    },
+    {
+      "epoch": 0.07924413288631514,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.940144848105379e-05,
+      "loss": 0.2267,
+      "step": 780
+    },
+    {
+      "epoch": 0.08026008330793458,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.93837877888615e-05,
+      "loss": 0.2597,
+      "step": 790
+    },
+    {
+      "epoch": 0.081276033729554,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.9365873574971745e-05,
+      "loss": 0.3701,
+      "step": 800
+    },
+    {
+      "epoch": 0.08229198415117342,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.9347706025641136e-05,
+      "loss": 0.2559,
+      "step": 810
+    },
+    {
+      "epoch": 0.08330793457279284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.9329285329760275e-05,
+      "loss": 0.2799,
+      "step": 820
+    },
+    {
+      "epoch": 0.08432388499441228,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.9310611678851735e-05,
+      "loss": 0.2866,
+      "step": 830
+    },
+    {
+      "epoch": 0.0853398354160317,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.929168526706811e-05,
+      "loss": 0.3105,
+      "step": 840
+    },
+    {
+      "epoch": 0.08635578583765112,
+      "grad_norm": 13.625,
+      "learning_rate": 4.927250629119e-05,
+      "loss": 0.2454,
+      "step": 850
+    },
+    {
+      "epoch": 0.08737173625927054,
+      "grad_norm": 3.921875,
+      "learning_rate": 4.9253074950623925e-05,
+      "loss": 0.2424,
+      "step": 860
+    },
+    {
+      "epoch": 0.08838768668088998,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.9233391447400286e-05,
+      "loss": 0.2481,
+      "step": 870
+    },
+    {
+      "epoch": 0.0894036371025094,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.921345598617125e-05,
+      "loss": 0.2231,
+      "step": 880
+    },
+    {
+      "epoch": 0.09041958752412882,
+      "grad_norm": 5.375,
+      "learning_rate": 4.9193268774208654e-05,
+      "loss": 0.3447,
+      "step": 890
+    },
+    {
+      "epoch": 0.09143553794574824,
+      "grad_norm": 2.0,
+      "learning_rate": 4.9172830021401785e-05,
+      "loss": 0.229,
+      "step": 900
+    },
+    {
+      "epoch": 0.09245148836736768,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.9152139940255245e-05,
+      "loss": 0.2122,
+      "step": 910
+    },
+    {
+      "epoch": 0.0934674387889871,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.913119874588677e-05,
+      "loss": 0.2386,
+      "step": 920
+    },
+    {
+      "epoch": 0.09448338921060652,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.911000665602489e-05,
+      "loss": 0.1944,
+      "step": 930
+    },
+    {
+      "epoch": 0.09549933963222594,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.9088563891006786e-05,
+      "loss": 0.2038,
+      "step": 940
+    },
+    {
+      "epoch": 0.09651529005384538,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.906687067377592e-05,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.0975312404754648,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.904492722987976e-05,
+      "loss": 0.3157,
+      "step": 960
+    },
+    {
+      "epoch": 0.09854719089708422,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.902273378746738e-05,
+      "loss": 0.3077,
+      "step": 970
+    },
+    {
+      "epoch": 0.09956314131870364,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.9000290577287165e-05,
+      "loss": 0.2756,
+      "step": 980
+    },
+    {
+      "epoch": 0.10057909174032308,
+      "grad_norm": 0.99609375,
+      "learning_rate": 4.897759783268434e-05,
+      "loss": 0.2915,
+      "step": 990
+    },
+    {
+      "epoch": 0.1015950421619425,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.895465578959859e-05,
+      "loss": 0.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10261099258356192,
+      "grad_norm": 4.0,
+      "learning_rate": 4.893146468656159e-05,
+      "loss": 0.2499,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.890802476469452e-05,
+      "loss": 0.278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10464289342680078,
+      "grad_norm": 3.625,
+      "learning_rate": 4.888433626770558e-05,
+      "loss": 0.2143,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1056588438484202,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.886039944188741e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.10667479427003962,
+      "grad_norm": 4.5,
+      "learning_rate": 4.883621453611461e-05,
+      "loss": 0.2744,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10769074469165904,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.881178180184106e-05,
+      "loss": 0.2734,
+      "step": 1060
+    },
+    {
+      "epoch": 0.10870669511327848,
+      "grad_norm": 3.125,
+      "learning_rate": 4.878710149309735e-05,
+      "loss": 0.3574,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1097226455348979,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.876217386648816e-05,
+      "loss": 0.2625,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11073859595651732,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.873699918118955e-05,
+      "loss": 0.2437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11175454637813674,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.87115776989463e-05,
+      "loss": 0.2051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11277049679975618,
+      "grad_norm": 4.375,
+      "learning_rate": 4.8685909684069153e-05,
+      "loss": 0.1727,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1137864472213756,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.865999540343211e-05,
+      "loss": 0.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11480239764299502,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.86338351264696e-05,
+      "loss": 0.3529,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11581834806461444,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.8607429125173754e-05,
+      "loss": 0.2113,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11683429848623388,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.858077767409149e-05,
+      "loss": 0.2759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1178502489078533,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.855388105032174e-05,
+      "loss": 0.2482,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11886619932947272,
+      "grad_norm": 3.5,
+      "learning_rate": 4.852673953351249e-05,
+      "loss": 0.1865,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11988214975109214,
+      "grad_norm": 3.75,
+      "learning_rate": 4.849935340585796e-05,
+      "loss": 0.2659,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12089810017271158,
+      "grad_norm": 3.375,
+      "learning_rate": 4.8471722952095586e-05,
+      "loss": 0.1506,
+      "step": 1190
+    },
+    {
+      "epoch": 0.121914050594331,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.844384845950312e-05,
+      "loss": 0.307,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12293000101595042,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.841573021789561e-05,
+      "loss": 0.1952,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12394595143756984,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.838736851962239e-05,
+      "loss": 0.1779,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12496190185918928,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.835876365956408e-05,
+      "loss": 0.1235,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12597785228080868,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8329915935129436e-05,
+      "loss": 0.1876,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12699380270242813,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.830082564625235e-05,
+      "loss": 0.2188,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12800975312404755,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.8271493095388684e-05,
+      "loss": 0.2622,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12902570354566698,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.824191858751312e-05,
+      "loss": 0.2724,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1300416539672864,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.821210243011601e-05,
+      "loss": 0.2413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13105760438890582,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.818204493320016e-05,
+      "loss": 0.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13207355481052524,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.8151746409277634e-05,
+      "loss": 0.2295,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13308950523214466,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.8121207173366484e-05,
+      "loss": 0.2733,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13410545565376408,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.809042754298746e-05,
+      "loss": 0.2311,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13512140607538353,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.805940783816075e-05,
+      "loss": 0.2059,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13613735649700295,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.8028148381402625e-05,
+      "loss": 0.2102,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13715330691862238,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.7996649497722084e-05,
+      "loss": 0.2708,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1381692573402418,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7964911514617485e-05,
+      "loss": 0.2429,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13918520776186122,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.793293476207312e-05,
+      "loss": 0.2725,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14020115818348064,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.790071957255585e-05,
+      "loss": 0.2098,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14121710860510006,
+      "grad_norm": 4.25,
+      "learning_rate": 4.786826628101154e-05,
+      "loss": 0.2101,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14223305902671948,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.783557522486167e-05,
+      "loss": 0.2624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14324900944833893,
+      "grad_norm": 3.125,
+      "learning_rate": 4.780264674399978e-05,
+      "loss": 0.2518,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14426495986995835,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.7769481180787966e-05,
+      "loss": 0.3112,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14528091029157778,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.773607888005327e-05,
+      "loss": 0.2747,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1462968607131972,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.770244018908416e-05,
+      "loss": 0.1572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14731281113481662,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.766856545762687e-05,
+      "loss": 0.2148,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14832876155643604,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.763445503788178e-05,
+      "loss": 0.2531,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14934471197805546,
+      "grad_norm": 2.375,
+      "learning_rate": 4.760010928449976e-05,
+      "loss": 0.199,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15036066239967488,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.7565528554578485e-05,
+      "loss": 0.2366,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15137661282129433,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.75307132076587e-05,
+      "loss": 0.1862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15239256324291375,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.749566360572049e-05,
+      "loss": 0.2143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15340851366453317,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.746038011317955e-05,
+      "loss": 0.1877,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1544244640861526,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.742486309688333e-05,
+      "loss": 0.2831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15544041450777202,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.738911292610732e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15645636492939144,
+      "grad_norm": 3.953125,
+      "learning_rate": 4.735312997255107e-05,
+      "loss": 0.192,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15747231535101086,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.7316914610334475e-05,
+      "loss": 0.2586,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15848826577263028,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.728046721599378e-05,
+      "loss": 0.2141,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15950421619424973,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.724378816847771e-05,
+      "loss": 0.193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16052016661586915,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.720687784914352e-05,
+      "loss": 0.191,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16153611703748857,
+      "grad_norm": 3.75,
+      "learning_rate": 4.716973664175304e-05,
+      "loss": 0.2172,
+      "step": 1590
+    },
+    {
+      "epoch": 0.162552067459108,
+      "grad_norm": 3.125,
+      "learning_rate": 4.7132364932468645e-05,
+      "loss": 0.2134,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16356801788072742,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.709476310984932e-05,
+      "loss": 0.2055,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16458396830234684,
+      "grad_norm": 3.875,
+      "learning_rate": 4.705693156484652e-05,
+      "loss": 0.2136,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16559991872396626,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.7018870690800196e-05,
+      "loss": 0.1471,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16661586914558568,
+      "grad_norm": 2.5,
+      "learning_rate": 4.698058088343465e-05,
+      "loss": 0.2308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16763181956720513,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6942062540854425e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16864776998882455,
+      "grad_norm": 3.125,
+      "learning_rate": 4.69033160635402e-05,
+      "loss": 0.2654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16966372041044397,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.6864341854344587e-05,
+      "loss": 0.2226,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1706796708320634,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.682514031848795e-05,
+      "loss": 0.2438,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17169562125368282,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.678571186355423e-05,
+      "loss": 0.1889,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17271157167530224,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.6746056899486644e-05,
+      "loss": 0.2117,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17372752209692166,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.67061758385835e-05,
+      "loss": 0.1953,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17474347251854108,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.6666069095493816e-05,
+      "loss": 0.1844,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17575942294016053,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.662573708721309e-05,
+      "loss": 0.2774,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17677537336177995,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.658518023307894e-05,
+      "loss": 0.2527,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17779132378339937,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.654439895476671e-05,
+      "loss": 0.2164,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1788072742050188,
+      "grad_norm": 2.390625,
+      "learning_rate": 4.6503393676285146e-05,
+      "loss": 0.2424,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17982322462663822,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.646216482397192e-05,
+      "loss": 0.2428,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18083917504825764,
+      "grad_norm": 2.796875,
+      "learning_rate": 4.6420712826489275e-05,
+      "loss": 0.2155,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18185512546987706,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.6379038114819485e-05,
+      "loss": 0.1544,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18287107589149648,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.6337141122260444e-05,
+      "loss": 0.2029,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18388702631311593,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.629502228442112e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18490297673473535,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.6252682039217045e-05,
+      "loss": 0.2101,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18591892715635477,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.621012082686573e-05,
+      "loss": 0.2076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1869348775779742,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.616733908988216e-05,
+      "loss": 0.2719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18795082799959362,
+      "grad_norm": 1.953125,
+      "learning_rate": 4.612433727307409e-05,
+      "loss": 0.2105,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18896677842121304,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.608111582353751e-05,
+      "loss": 0.1877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18998272884283246,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.603767519065197e-05,
+      "loss": 0.2238,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19099867926445188,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.599401582607589e-05,
+      "loss": 0.243,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19201462968607133,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.595013818374185e-05,
+      "loss": 0.1867,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19303058010769075,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.5906042719851925e-05,
+      "loss": 0.1994,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19404653052931017,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.586172989287291e-05,
+      "loss": 0.1899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1950624809509296,
+      "grad_norm": 2.6875,
+      "learning_rate": 4.5817200163531534e-05,
+      "loss": 0.2528,
+      "step": 1920
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.577245399480972e-05,
+      "loss": 0.2336,
+      "step": 1930
+    },
+    {
+      "epoch": 0.19709438179416844,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.5727491851939715e-05,
+      "loss": 0.2204,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19811033221578786,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.568231420239929e-05,
+      "loss": 0.1656,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19912628263740728,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.563692151590687e-05,
+      "loss": 0.2105,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20014223305902673,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5591314264416666e-05,
+      "loss": 0.1464,
+      "step": 1970
+    },
+    {
+      "epoch": 0.20115818348064615,
+      "grad_norm": 4.25,
+      "learning_rate": 4.554549292211371e-05,
+      "loss": 0.2103,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20217413390226557,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.549945796540901e-05,
+      "loss": 0.144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.203190084323885,
+      "grad_norm": 1.859375,
+      "learning_rate": 4.545320987293453e-05,
+      "loss": 0.1963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.20420603474550442,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.540674912553824e-05,
+      "loss": 0.2115,
+      "step": 2010
+    },
+    {
+      "epoch": 0.20522198516712384,
+      "grad_norm": 4.25,
+      "learning_rate": 4.536007620627911e-05,
+      "loss": 0.1682,
+      "step": 2020
+    },
+    {
+      "epoch": 0.20623793558874326,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.531319160042212e-05,
+      "loss": 0.1992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.5266095795433126e-05,
+      "loss": 0.1134,
+      "step": 2040
+    },
+    {
+      "epoch": 0.20826983643198213,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.5218789280973925e-05,
+      "loss": 0.1474,
+      "step": 2050
+    },
+    {
+      "epoch": 0.20928578685360155,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.5171272548897024e-05,
+      "loss": 0.1955,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21030173727522097,
+      "grad_norm": 2.734375,
+      "learning_rate": 4.512354609324063e-05,
+      "loss": 0.2042,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2113176876968404,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.507561041022347e-05,
+      "loss": 0.2174,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21233363811845982,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.502746599823963e-05,
+      "loss": 0.2634,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21334958854007924,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.497911335785339e-05,
+      "loss": 0.1884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21436553896169866,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.4930552991794e-05,
+      "loss": 0.1872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21538148938331808,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.4881785404950474e-05,
+      "loss": 0.2233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.21639743980493753,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.483281110436631e-05,
+      "loss": 0.2374,
+      "step": 2130
+    },
+    {
+      "epoch": 0.21741339022655695,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.478363059923426e-05,
+      "loss": 0.2545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.21842934064817637,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.4734244400891014e-05,
+      "loss": 0.2063,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2194452910697958,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.4684653022811865e-05,
+      "loss": 0.1219,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22046124149141522,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.463485698060541e-05,
+      "loss": 0.2805,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22147719191303464,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.458485679200814e-05,
+      "loss": 0.1998,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22249314233465406,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.453465297687912e-05,
+      "loss": 0.2489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22350909275627348,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.448424605719452e-05,
+      "loss": 0.2731,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22452504317789293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.443363655704224e-05,
+      "loss": 0.2425,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22554099359951235,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.438282500261641e-05,
+      "loss": 0.2938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.22655694402113177,
+      "grad_norm": 1.1953125,
+      "learning_rate": 4.433181192221197e-05,
+      "loss": 0.1728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2275728944427512,
+      "grad_norm": 1.34375,
+      "learning_rate": 4.4280597846219155e-05,
+      "loss": 0.216,
+      "step": 2240
+    },
+    {
+      "epoch": 0.22858884486437062,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.422918330711796e-05,
+      "loss": 0.1612,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22960479528599004,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.417756883947263e-05,
+      "loss": 0.107,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23062074570760946,
+      "grad_norm": 3.375,
+      "learning_rate": 4.412575497992611e-05,
+      "loss": 0.1756,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23163669612922888,
+      "grad_norm": 4.375,
+      "learning_rate": 4.407374226719445e-05,
+      "loss": 0.234,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23265264655084833,
+      "grad_norm": 3.25,
+      "learning_rate": 4.402153124206119e-05,
+      "loss": 0.2144,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23366859697246775,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.396912244737173e-05,
+      "loss": 0.1696,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23468454739408717,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.391651642802778e-05,
+      "loss": 0.2506,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2357004978157066,
+      "grad_norm": 4.5,
+      "learning_rate": 4.386371373098155e-05,
+      "loss": 0.1686,
+      "step": 2320
+    },
+    {
+      "epoch": 0.23671644823732602,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.381071490523018e-05,
+      "loss": 0.2403,
+      "step": 2330
+    },
+    {
+      "epoch": 0.23773239865894544,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3757520501809955e-05,
+      "loss": 0.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.23874834908056486,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.370413107379065e-05,
+      "loss": 0.1698,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23976429950218428,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.36505471762697e-05,
+      "loss": 0.1928,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24078024992380373,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.3596769366366474e-05,
+      "loss": 0.2035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24179620034542315,
+      "grad_norm": 5.75,
+      "learning_rate": 4.354279820321649e-05,
+      "loss": 0.16,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24281215076704257,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.34886342479656e-05,
+      "loss": 0.1851,
+      "step": 2390
+    },
+    {
+      "epoch": 0.243828101188662,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.34342780637641e-05,
+      "loss": 0.1726,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24484405161028142,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.337973021576095e-05,
+      "loss": 0.2847,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24586000203190084,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.3324991271097846e-05,
+      "loss": 0.2528,
+      "step": 2420
+    },
+    {
+      "epoch": 0.24687595245352026,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.3270061798903374e-05,
+      "loss": 0.1573,
+      "step": 2430
+    },
+    {
+      "epoch": 0.24789190287513968,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.321494237028701e-05,
+      "loss": 0.1703,
+      "step": 2440
+    },
+    {
+      "epoch": 0.24890785329675913,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.31596335583333e-05,
+      "loss": 0.2613,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24992380371837855,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.310413593809579e-05,
+      "loss": 0.22,
+      "step": 2460
+    },
+    {
+      "epoch": 0.250939754139998,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.304845008659108e-05,
+      "loss": 0.1263,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25195570456161737,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2992576582792895e-05,
+      "loss": 0.1639,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2529716549832368,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.293651600762595e-05,
+      "loss": 0.2681,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25398760540485626,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.288026894395999e-05,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25500355582647566,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.2823835976603723e-05,
+      "loss": 0.2324,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2560195062480951,
+      "grad_norm": 5.625,
+      "learning_rate": 4.276721769229869e-05,
+      "loss": 0.1834,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2570354566697145,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.271041467971323e-05,
+      "loss": 0.1826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.25805140709133395,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.265342752943632e-05,
+      "loss": 0.2463,
+      "step": 2540
+    },
+    {
+      "epoch": 0.25906735751295334,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2596256833971425e-05,
+      "loss": 0.2598,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2600833079345728,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.2538903187730374e-05,
+      "loss": 0.1148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26109925835619224,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.248136718702716e-05,
+      "loss": 0.2123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26211520877781164,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.242364943007172e-05,
+      "loss": 0.2369,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2631311591994311,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.236575051696377e-05,
+      "loss": 0.261,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2641471096210505,
+      "grad_norm": 2.75,
+      "learning_rate": 4.2307671049686514e-05,
+      "loss": 0.1564,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26516306004266993,
+      "grad_norm": 3.5,
+      "learning_rate": 4.2249411632100396e-05,
+      "loss": 0.1563,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2661790104642893,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.219097286993684e-05,
+      "loss": 0.1697,
+      "step": 2620
+    },
+    {
+      "epoch": 0.26719496088590877,
+      "grad_norm": 2.125,
+      "learning_rate": 4.2132355370791946e-05,
+      "loss": 0.1844,
+      "step": 2630
+    },
+    {
+      "epoch": 0.26821091130752817,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.2073559744120156e-05,
+      "loss": 0.2144,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2692268617291476,
+      "grad_norm": 2.375,
+      "learning_rate": 4.201458660122793e-05,
+      "loss": 0.2013,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27024281215076706,
+      "grad_norm": 3.625,
+      "learning_rate": 4.1955436555267393e-05,
+      "loss": 0.2166,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27125876257238646,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.189611022122997e-05,
+      "loss": 0.1934,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2722747129940059,
+      "grad_norm": 2.75,
+      "learning_rate": 4.1836608215939944e-05,
+      "loss": 0.2157,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2732906634156253,
+      "grad_norm": 3.5,
+      "learning_rate": 4.17769311580481e-05,
+      "loss": 0.18,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27430661383724475,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.171707966802528e-05,
+      "loss": 0.2178,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27532256425886414,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.16570543681559e-05,
+      "loss": 0.1896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2763385146804836,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.159685588253151e-05,
+      "loss": 0.1322,
+      "step": 2720
+    },
+    {
+      "epoch": 0.27735446510210304,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.153648483704429e-05,
+      "loss": 0.184,
+      "step": 2730
+    },
+    {
+      "epoch": 0.27837041552372244,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.147594185938057e-05,
+      "loss": 0.2451,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2793863659453419,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.141522757901426e-05,
+      "loss": 0.2367,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2804023163669613,
+      "grad_norm": 3.375,
+      "learning_rate": 4.1354342627200345e-05,
+      "loss": 0.179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28141826678858073,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.1293287636968286e-05,
+      "loss": 0.1396,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2824342172102001,
+      "grad_norm": 2.546875,
+      "learning_rate": 4.1232063243115485e-05,
+      "loss": 0.1963,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28345016763181957,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.117067008220063e-05,
+      "loss": 0.2457,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28446611805343897,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.110910879253712e-05,
+      "loss": 0.2262,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2854820684750584,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.104738001418641e-05,
+      "loss": 0.2499,
+      "step": 2810
+    },
+    {
+      "epoch": 0.28649801889667786,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.098548438895135e-05,
+      "loss": 0.1667,
+      "step": 2820
+    },
+    {
+      "epoch": 0.28751396931829726,
+      "grad_norm": 2.875,
+      "learning_rate": 4.092342256036954e-05,
+      "loss": 0.2288,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2885299197399167,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.086119517370659e-05,
+      "loss": 0.2038,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2895458701615361,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.0798802875949485e-05,
+      "loss": 0.181,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29056182058315555,
+      "grad_norm": 2.296875,
+      "learning_rate": 4.073624631579975e-05,
+      "loss": 0.1886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29157777100477494,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.067352614366685e-05,
+      "loss": 0.2053,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2925937214263944,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.061064301166128e-05,
+      "loss": 0.1409,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29360967184801384,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.054759757358787e-05,
+      "loss": 0.184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29462562226963324,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.048439048493898e-05,
+      "loss": 0.2306,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2956415726912527,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.0421022402887676e-05,
+      "loss": 0.1914,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2966575231128721,
+      "grad_norm": 2.3125,
+      "learning_rate": 4.035749398628088e-05,
+      "loss": 0.1653,
+      "step": 2920
+    },
+    {
+      "epoch": 0.29767347353449153,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.029380589563256e-05,
+      "loss": 0.1941,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2986894239561109,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.02299587931168e-05,
+      "loss": 0.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.29970537437773037,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.0165953342560974e-05,
+      "loss": 0.1605,
+      "step": 2950
+    },
+    {
+      "epoch": 0.30072132479934977,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.010179020943884e-05,
+      "loss": 0.1726,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3017372752209692,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.003747006086357e-05,
+      "loss": 0.2208,
+      "step": 2970
+    },
+    {
+      "epoch": 0.30275322564258866,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.9972993565580866e-05,
+      "loss": 0.1325,
+      "step": 2980
+    },
+    {
+      "epoch": 0.30376917606420806,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.9908361393962e-05,
+      "loss": 0.2014,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3047851264858275,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.984357421799681e-05,
+      "loss": 0.165,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3058010769074469,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.9778632711286756e-05,
+      "loss": 0.212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.30681702732906635,
+      "grad_norm": 4.25,
+      "learning_rate": 3.971353754903788e-05,
+      "loss": 0.2388,
+      "step": 3020
+    },
+    {
+      "epoch": 0.30783297775068574,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.964828940805381e-05,
+      "loss": 0.2175,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3088489281723052,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.95828889667287e-05,
+      "loss": 0.2088,
+      "step": 3040
+    },
+    {
+      "epoch": 0.30986487859392464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.9517336905040244e-05,
+      "loss": 0.1913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.9451633904542483e-05,
+      "loss": 0.2185,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3118967794371635,
+      "grad_norm": 2.59375,
+      "learning_rate": 3.9385780648358846e-05,
+      "loss": 0.2072,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3129127298587829,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.9319777821174955e-05,
+      "loss": 0.1902,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31392868028040233,
+      "grad_norm": 2.375,
+      "learning_rate": 3.925362610923158e-05,
+      "loss": 0.259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3149446307020217,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.918732620031742e-05,
+      "loss": 0.2026,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31596058112364117,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.912087878376205e-05,
+      "loss": 0.1478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.31697653154526056,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.905428455042865e-05,
+      "loss": 0.167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.31799248196688,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.898754419270693e-05,
+      "loss": 0.1629,
+      "step": 3130
+    },
+    {
+      "epoch": 0.31900843238849946,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.892065840450583e-05,
+      "loss": 0.1308,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32002438281011886,
+      "grad_norm": 4.625,
+      "learning_rate": 3.885362788124637e-05,
+      "loss": 0.2008,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3210403332317383,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.8786453319854396e-05,
+      "loss": 0.2225,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3220562836533577,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.8719135418753366e-05,
+      "loss": 0.2243,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32307223407497715,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.865167487785702e-05,
+      "loss": 0.1981,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32408818449659654,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.8584072398562164e-05,
+      "loss": 0.2031,
+      "step": 3190
+    },
+    {
+      "epoch": 0.325104134918216,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.851632868374136e-05,
+      "loss": 0.1621,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32612008533983544,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.844844443773562e-05,
+      "loss": 0.1674,
+      "step": 3210
+    },
+    {
+      "epoch": 0.32713603576145484,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.8380420366347046e-05,
+      "loss": 0.1502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3281519861830743,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.831225717683157e-05,
+      "loss": 0.1868,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3291679366046937,
+      "grad_norm": 2.703125,
+      "learning_rate": 3.8243955577891534e-05,
+      "loss": 0.1818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3301838870263131,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.8175516279668335e-05,
+      "loss": 0.2215,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3311998374479325,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.810693999373505e-05,
+      "loss": 0.2544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33221578786955197,
+      "grad_norm": 4.0,
+      "learning_rate": 3.8038227433089056e-05,
+      "loss": 0.1175,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33323173829117136,
+      "grad_norm": 3.625,
+      "learning_rate": 3.796937931214458e-05,
+      "loss": 0.2213,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3342476887127908,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7900396346725296e-05,
+      "loss": 0.1711,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33526363913441026,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.783127925405686e-05,
+      "loss": 0.2628,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33627958955602966,
+      "grad_norm": 2.1875,
+      "learning_rate": 3.77620287527595e-05,
+      "loss": 0.1671,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3372955399776491,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.769264556284048e-05,
+      "loss": 0.2109,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3383114903992685,
+      "grad_norm": 2.875,
+      "learning_rate": 3.762313040568665e-05,
+      "loss": 0.1978,
+      "step": 3330
+    },
+    {
+      "epoch": 0.33932744082088795,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.755348400405697e-05,
+      "loss": 0.1275,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34034339124250734,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.7483707082074945e-05,
+      "loss": 0.1482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3413593416641268,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.741380036522111e-05,
+      "loss": 0.1933,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34237529208574624,
+      "grad_norm": 4.53125,
+      "learning_rate": 3.734376458032551e-05,
+      "loss": 0.1925,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34339124250736563,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.727360045556014e-05,
+      "loss": 0.2297,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3444071929289851,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.7203308720431336e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3454231433506045,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.7132890105772234e-05,
+      "loss": 0.258,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3464390937722239,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.706234534373515e-05,
+      "loss": 0.2376,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3474550441938433,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.6991675167783985e-05,
+      "loss": 0.2403,
+      "step": 3420
+    },
+    {
+      "epoch": 0.34847099461546277,
+      "grad_norm": 1.1640625,
+      "learning_rate": 3.6920880312686556e-05,
+      "loss": 0.1642,
+      "step": 3430
+    },
+    {
+      "epoch": 0.34948694503708216,
+      "grad_norm": 2.875,
+      "learning_rate": 3.684996151450702e-05,
+      "loss": 0.1455,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3505028954587016,
+      "grad_norm": 0.59765625,
+      "learning_rate": 3.6778919510598155e-05,
+      "loss": 0.2175,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35151884588032106,
+      "grad_norm": 0.93359375,
+      "learning_rate": 3.670775503959376e-05,
+      "loss": 0.1858,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35253479630194046,
+      "grad_norm": 4.1875,
+      "learning_rate": 3.6636468841400917e-05,
+      "loss": 0.1911,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3535507467235599,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.656506165719233e-05,
+      "loss": 0.2114,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3545666971451793,
+      "grad_norm": 1.171875,
+      "learning_rate": 3.649353422939863e-05,
+      "loss": 0.1841,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35558264756679875,
+      "grad_norm": 2.53125,
+      "learning_rate": 3.6421887301700615e-05,
+      "loss": 0.1505,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35659859798841814,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.6350121619021524e-05,
+      "loss": 0.2625,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3576145484100376,
+      "grad_norm": 5.25,
+      "learning_rate": 3.627823792751936e-05,
+      "loss": 0.1676,
+      "step": 3520
+    },
+    {
+      "epoch": 0.35863049883165704,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.620623697457905e-05,
+      "loss": 0.1963,
+      "step": 3530
+    },
+    {
+      "epoch": 0.35964644925327643,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.613411950880468e-05,
+      "loss": 0.2048,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3606623996748959,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.606188628001178e-05,
+      "loss": 0.226,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3616783500965153,
+      "grad_norm": 2.375,
+      "learning_rate": 3.598953803921947e-05,
+      "loss": 0.1884,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3626943005181347,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.591707553864266e-05,
+      "loss": 0.224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3637102509397541,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.584449953168423e-05,
+      "loss": 0.1866,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36472620136137357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.577181077292722e-05,
+      "loss": 0.1663,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36574215178299296,
+      "grad_norm": 5.0,
+      "learning_rate": 3.569901001812696e-05,
+      "loss": 0.2032,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3667581022046124,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.562609802420321e-05,
+      "loss": 0.2395,
+      "step": 3610
+    },
+    {
+      "epoch": 0.36777405262623186,
+      "grad_norm": 3.796875,
+      "learning_rate": 3.555307554923229e-05,
+      "loss": 0.1799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.36879000304785126,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.547994335243925e-05,
+      "loss": 0.1771,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3698059534694707,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.540670219418989e-05,
+      "loss": 0.2123,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3708219038910901,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.53333528359829e-05,
+      "loss": 0.2159,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37183785431270955,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.525989604044198e-05,
+      "loss": 0.2749,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37285380473432894,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5186332571307826e-05,
+      "loss": 0.1613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3738697551559484,
+      "grad_norm": 3.984375,
+      "learning_rate": 3.511266319343025e-05,
+      "loss": 0.1877,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37488570557756784,
+      "grad_norm": 2.203125,
+      "learning_rate": 3.503888867276022e-05,
+      "loss": 0.2185,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37590165599918723,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.4965009776341894e-05,
+      "loss": 0.2195,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3769176064208067,
+      "grad_norm": 4.375,
+      "learning_rate": 3.489102727230461e-05,
+      "loss": 0.2344,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3779335568424261,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.481694192985496e-05,
+      "loss": 0.1863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3789495072640455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 3.474275451926875e-05,
+      "loss": 0.1894,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3799654576856649,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.4668465811883e-05,
+      "loss": 0.2127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38098140810728437,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.4594076580087914e-05,
+      "loss": 0.2125,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38199735852890376,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.451958759731889e-05,
+      "loss": 0.1801,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3830133089505232,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.4444999638048456e-05,
+      "loss": 0.1949,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38402925937214266,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.437031347777817e-05,
+      "loss": 0.2719,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38504520979376206,
+      "grad_norm": 3.9375,
+      "learning_rate": 3.4295529893030634e-05,
+      "loss": 0.1697,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3860611602153815,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.422064966134138e-05,
+      "loss": 0.1557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3870771106370009,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.4145673561250794e-05,
+      "loss": 0.2129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.38809306105862035,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.4070602372296e-05,
+      "loss": 0.2068,
+      "step": 3820
+    },
+    {
+      "epoch": 0.38910901148023974,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.39954368750028e-05,
+      "loss": 0.1634,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3901249619018592,
+      "grad_norm": 1.75,
+      "learning_rate": 3.392017785087752e-05,
+      "loss": 0.2299,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39114091232347864,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.38448260823989e-05,
+      "loss": 0.1585,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.8125,
+      "learning_rate": 3.376938235300996e-05,
+      "loss": 0.2382,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3931728131667175,
+      "grad_norm": 5.375,
+      "learning_rate": 3.369384744710984e-05,
+      "loss": 0.1987,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3941887635883369,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.361822215004566e-05,
+      "loss": 0.2316,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3952047140099563,
+      "grad_norm": 2.0,
+      "learning_rate": 3.354250724810436e-05,
+      "loss": 0.2019,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3962206644315757,
+      "grad_norm": 2.3125,
+      "learning_rate": 3.34667035285045e-05,
+      "loss": 0.187,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39723661485319517,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.339081177938811e-05,
+      "loss": 0.2353,
+      "step": 3910
+    },
+    {
+      "epoch": 0.39825256527481456,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.331483278981244e-05,
+      "loss": 0.2078,
+      "step": 3920
+    },
+    {
+      "epoch": 0.399268515696434,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.323876734974183e-05,
+      "loss": 0.1761,
+      "step": 3930
+    },
+    {
+      "epoch": 0.40028446611805346,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.316261625003943e-05,
+      "loss": 0.2081,
+      "step": 3940
+    },
+    {
+      "epoch": 0.40130041653967286,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.308638028245902e-05,
+      "loss": 0.2087,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4023163669612923,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.301006023963676e-05,
+      "loss": 0.1579,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4033323173829117,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.293365691508295e-05,
+      "loss": 0.1904,
+      "step": 3970
+    },
+    {
+      "epoch": 0.40434826780453115,
+      "grad_norm": 3.0,
+      "learning_rate": 3.285717110317379e-05,
+      "loss": 0.1991,
+      "step": 3980
+    },
+    {
+      "epoch": 0.40536421822615054,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.27806035991431e-05,
+      "loss": 0.1445,
+      "step": 3990
+    },
+    {
+      "epoch": 0.40638016864777,
+      "grad_norm": 1.0859375,
+      "learning_rate": 3.2703955199074075e-05,
+      "loss": 0.2393,
+      "step": 4000
+    },
+    {
+      "epoch": 0.40739611906938944,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.262722669989098e-05,
+      "loss": 0.1789,
+      "step": 4010
+    },
+    {
+      "epoch": 0.40841206949100883,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.255041889935092e-05,
+      "loss": 0.1511,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4094280199126283,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.247353259603547e-05,
+      "loss": 0.2066,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4104439703342477,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.239656858934242e-05,
+      "loss": 0.1564,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4114599207558671,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.231952767947746e-05,
+      "loss": 0.1503,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4124758711774865,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.2242410667445844e-05,
+      "loss": 0.1633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41349182159910597,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.2165218355044076e-05,
+      "loss": 0.1492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2087951544851566e-05,
+      "loss": 0.3051,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4155237224423448,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.20106110402223e-05,
+      "loss": 0.2229,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41653967286396426,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.1933197645276455e-05,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41755562328558365,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.185571216489209e-05,
+      "loss": 0.1297,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4185715737072031,
+      "grad_norm": 3.625,
+      "learning_rate": 3.177815540469669e-05,
+      "loss": 0.2074,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4195875241288225,
+      "grad_norm": 2.296875,
+      "learning_rate": 3.1700528171058916e-05,
+      "loss": 0.1949,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42060347455044195,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.162283127108011e-05,
+      "loss": 0.1661,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42161942497206134,
+      "grad_norm": 2.5,
+      "learning_rate": 3.154506551258594e-05,
+      "loss": 0.2275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4226353753936808,
+      "grad_norm": 2.96875,
+      "learning_rate": 3.146723170411804e-05,
+      "loss": 0.2242,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42365132581530024,
+      "grad_norm": 6.625,
+      "learning_rate": 3.138933065492552e-05,
+      "loss": 0.1897,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42466727623691963,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.131136317495665e-05,
+      "loss": 0.1629,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4256832266585391,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.1233330074850364e-05,
+      "loss": 0.1535,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4266991770801585,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.115523216592786e-05,
+      "loss": 0.2494,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4277151275017779,
+      "grad_norm": 2.578125,
+      "learning_rate": 3.107707026018417e-05,
+      "loss": 0.1705,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4287310779233973,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.09988451702797e-05,
+      "loss": 0.1507,
+      "step": 4220
+    },
+    {
+      "epoch": 0.42974702834501677,
+      "grad_norm": 2.421875,
+      "learning_rate": 3.0920557709531804e-05,
+      "loss": 0.3071,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43076297876663616,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.0842208691906306e-05,
+      "loss": 0.199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4317789291882556,
+      "grad_norm": 3.5625,
+      "learning_rate": 3.076379893200904e-05,
+      "loss": 0.1987,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43279487960987506,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.068532924507739e-05,
+      "loss": 0.1945,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43381083003149445,
+      "grad_norm": 5.875,
+      "learning_rate": 3.060680044697183e-05,
+      "loss": 0.1937,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4348267804531139,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.052821335416739e-05,
+      "loss": 0.1643,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4358427308747333,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.0449568783745203e-05,
+      "loss": 0.1455,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43685868129635275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.0370867553384023e-05,
+      "loss": 0.1891,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43787463171797214,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.029211048135171e-05,
+      "loss": 0.1377,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4388905821395916,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.021329838649668e-05,
+      "loss": 0.2194,
+      "step": 4320
+    },
+    {
+      "epoch": 0.43990653256121104,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.0134432088239462e-05,
+      "loss": 0.1915,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44092248298283043,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.0055512406564146e-05,
+      "loss": 0.1794,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4419384334044499,
+      "grad_norm": 2.546875,
+      "learning_rate": 2.9976540162009836e-05,
+      "loss": 0.2154,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4429543838260693,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.9897516175662155e-05,
+      "loss": 0.1861,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4439703342476887,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.9818441269144693e-05,
+      "loss": 0.1857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4449862846693081,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9739316264610452e-05,
+      "loss": 0.1493,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44600223509092757,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.966014198473332e-05,
+      "loss": 0.186,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44701818551254696,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9580919252699502e-05,
+      "loss": 0.1963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4480341359341664,
+      "grad_norm": 7.3125,
+      "learning_rate": 2.9501648892198984e-05,
+      "loss": 0.2882,
+      "step": 4410
+    },
+    {
+      "epoch": 0.44905008635578586,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.942233172741693e-05,
+      "loss": 0.2154,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45006603677740525,
+      "grad_norm": 2.421875,
+      "learning_rate": 2.934296858302515e-05,
+      "loss": 0.2228,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4510819871990247,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9263560284173485e-05,
+      "loss": 0.1637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4520979376206441,
+      "grad_norm": 4.5,
+      "learning_rate": 2.91841076564813e-05,
+      "loss": 0.1396,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45311388804226355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.9104611526028808e-05,
+      "loss": 0.186,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45412983846388294,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.902507271934855e-05,
+      "loss": 0.1706,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4551457888855024,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.8945492063416768e-05,
+      "loss": 0.2191,
+      "step": 4480
+    },
+    {
+      "epoch": 0.45616173930712184,
+      "grad_norm": 2.734375,
+      "learning_rate": 2.8865870385644823e-05,
+      "loss": 0.1651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45717768972874123,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.8786208513870583e-05,
+      "loss": 0.1907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4581936401503607,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.8706507276349815e-05,
+      "loss": 0.2256,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4592095905719801,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8626767501747588e-05,
+      "loss": 0.215,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4602255409935995,
+      "grad_norm": 2.296875,
+      "learning_rate": 2.854699001912964e-05,
+      "loss": 0.2241,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4612414914152189,
+      "grad_norm": 2.078125,
+      "learning_rate": 2.846717565795376e-05,
+      "loss": 0.1541,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46225744183683837,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.8387325248061164e-05,
+      "loss": 0.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46327339225845776,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.8307439619667897e-05,
+      "loss": 0.259,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4642893426800772,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.8227519603356157e-05,
+      "loss": 0.2205,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46530529310169666,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.8147566030065677e-05,
+      "loss": 0.2256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.46632124352331605,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8067579731085085e-05,
+      "loss": 0.1671,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4673371939449355,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7987561538043273e-05,
+      "loss": 0.2471,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4683531443665549,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.7907512282900727e-05,
+      "loss": 0.1749,
+      "step": 4610
+    },
+    {
+      "epoch": 0.46936909478817435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.782743279794091e-05,
+      "loss": 0.2276,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47038504520979374,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.7747323915761574e-05,
+      "loss": 0.1971,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4714009956314132,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.7667186469266122e-05,
+      "loss": 0.1951,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47241694605303264,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.7587021291654924e-05,
+      "loss": 0.2045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47343289647465203,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.750682921641672e-05,
+      "loss": 0.155,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4744488468962715,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7426611077319864e-05,
+      "loss": 0.2038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4754647973178909,
+      "grad_norm": 5.5,
+      "learning_rate": 2.734636770840372e-05,
+      "loss": 0.159,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4764807477395103,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7266099943969976e-05,
+      "loss": 0.1566,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4774966981611297,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.7185808618573943e-05,
+      "loss": 0.1927,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47851264858274917,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.710549456701592e-05,
+      "loss": 0.1873,
+      "step": 4710
+    },
+    {
+      "epoch": 0.47952859900436856,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.702515862433247e-05,
+      "loss": 0.2474,
+      "step": 4720
+    },
+    {
+      "epoch": 0.480544549425988,
+      "grad_norm": 1.1640625,
+      "learning_rate": 2.6944801625787795e-05,
+      "loss": 0.204,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48156049984760746,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.6864424406864984e-05,
+      "loss": 0.1758,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48257645026922685,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6784027803257377e-05,
+      "loss": 0.161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4835924006908463,
+      "grad_norm": 2.046875,
+      "learning_rate": 2.6703612650859848e-05,
+      "loss": 0.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4846083511124657,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.6623179785760148e-05,
+      "loss": 0.1858,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48562430153408515,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.6542730044230175e-05,
+      "loss": 0.176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.48664025195570454,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.6462264262717278e-05,
+      "loss": 0.1657,
+      "step": 4790
+    },
+    {
+      "epoch": 0.487656202377324,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6381783277835605e-05,
+      "loss": 0.2705,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48867215279894344,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.6301287926357355e-05,
+      "loss": 0.2252,
+      "step": 4810
+    },
+    {
+      "epoch": 0.48968810322056283,
+      "grad_norm": 0.734375,
+      "learning_rate": 2.622077904520411e-05,
+      "loss": 0.2141,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4907040536421823,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.6140257471438108e-05,
+      "loss": 0.1935,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4917200040638017,
+      "grad_norm": 3.625,
+      "learning_rate": 2.6059724042253574e-05,
+      "loss": 0.2121,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4927359544854211,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.5979179594967983e-05,
+      "loss": 0.1221,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4937519049070405,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.5898624967013367e-05,
+      "loss": 0.2208,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49476785532865997,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.5818060995927607e-05,
+      "loss": 0.1904,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49578380575027936,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.573748851934574e-05,
+      "loss": 0.1658,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4967997561718988,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5656908374991213e-05,
+      "loss": 0.1626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49781570659351826,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.557632140066721e-05,
+      "loss": 0.1905,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49883165701513765,
+      "grad_norm": 4.875,
+      "learning_rate": 2.5495728434247917e-05,
+      "loss": 0.2591,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4998476074367571,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.5415130313669845e-05,
+      "loss": 0.1359,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5008635578583766,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.5334527876923063e-05,
+      "loss": 0.2353,
+      "step": 4930
+    },
+    {
+      "epoch": 0.501879508279996,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.5253921962042525e-05,
+      "loss": 0.2173,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5028954587016153,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.5173313407099373e-05,
+      "loss": 0.1631,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5039114091232347,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.5092703050192163e-05,
+      "loss": 0.1884,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5049273595448542,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.501209172943819e-05,
+      "loss": 0.217,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5059433099664736,
+      "grad_norm": 4.375,
+      "learning_rate": 2.49314802829648e-05,
+      "loss": 0.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.506959260388093,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.4850869548900628e-05,
+      "loss": 0.2049,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5079752108097125,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.477026036536688e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5089911612313319,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.4689653570468677e-05,
+      "loss": 0.164,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5100071116529513,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.460905000228628e-05,
+      "loss": 0.1649,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5110230620745707,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4528450498866428e-05,
+      "loss": 0.1777,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5120390124961902,
+      "grad_norm": 3.0,
+      "learning_rate": 2.444785589821356e-05,
+      "loss": 0.1505,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5130549629178096,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.436726703828118e-05,
+      "loss": 0.2672,
+      "step": 5050
+    },
+    {
+      "epoch": 0.514070913339429,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.428668475696308e-05,
+      "loss": 0.1756,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5150868637610485,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.420610989208465e-05,
+      "loss": 0.1655,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5161028141826679,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.412554328139419e-05,
+      "loss": 0.1579,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5171187646042873,
+      "grad_norm": 2.28125,
+      "learning_rate": 2.404498576255416e-05,
+      "loss": 0.1599,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.6484375,
+      "learning_rate": 2.3964438173132522e-05,
+      "loss": 0.1508,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5191506654475262,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.388390135059395e-05,
+      "loss": 0.1578,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5201666158691456,
+      "grad_norm": 1.21875,
+      "learning_rate": 2.3803376132291226e-05,
+      "loss": 0.1374,
+      "step": 5120
+    },
+    {
+      "epoch": 0.521182566290765,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.3722863355456436e-05,
+      "loss": 0.1854,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5221985167123845,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.364236385719236e-05,
+      "loss": 0.1391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5232144671340039,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.356187847446366e-05,
+      "loss": 0.2106,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5242304175556233,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.348140804408829e-05,
+      "loss": 0.2383,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5252463679772427,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3400953402728713e-05,
+      "loss": 0.1537,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5262623183988622,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.332051538688322e-05,
+      "loss": 0.1841,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5272782688204816,
+      "grad_norm": 3.25,
+      "learning_rate": 2.3240094832877287e-05,
+      "loss": 0.1855,
+      "step": 5190
+    },
+    {
+      "epoch": 0.528294219242101,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.3159692576854793e-05,
+      "loss": 0.2625,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5293101696637205,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.3079309454769413e-05,
+      "loss": 0.1292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5303261200853399,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.2998946302375827e-05,
+      "loss": 0.1263,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5313420705069593,
+      "grad_norm": 2.71875,
+      "learning_rate": 2.2918603955221148e-05,
+      "loss": 0.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5323580209285786,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.283828324863613e-05,
+      "loss": 0.1231,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5333739713501982,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.2757985017726557e-05,
+      "loss": 0.1939,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5343899217718175,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.2677710097364495e-05,
+      "loss": 0.168,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5354058721934369,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.259745932217969e-05,
+      "loss": 0.1883,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5364218226150563,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.2517233526550817e-05,
+      "loss": 0.1898,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5374377730366758,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2437033544596837e-05,
+      "loss": 0.1838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5384537234582952,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.2356860210168336e-05,
+      "loss": 0.1553,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5394696738799146,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.2276714356838824e-05,
+      "loss": 0.2248,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5404856243015341,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.2196596817896118e-05,
+      "loss": 0.1421,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5415015747231535,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.2116508426333596e-05,
+      "loss": 0.1947,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5425175251447729,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.2036450014841652e-05,
+      "loss": 0.2207,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5435334755663923,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.19564224157989e-05,
+      "loss": 0.2208,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5445494259880118,
+      "grad_norm": 7.5625,
+      "learning_rate": 2.1876426461263654e-05,
+      "loss": 0.1739,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5455653764096312,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.179646298296519e-05,
+      "loss": 0.1938,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5465813268312506,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.171653281229511e-05,
+      "loss": 0.1736,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5475972772528701,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1636636780298732e-05,
+      "loss": 0.2167,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5486132276744895,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1556775717666427e-05,
+      "loss": 0.1711,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5496291780961089,
+      "grad_norm": 5.125,
+      "learning_rate": 2.147695045472499e-05,
+      "loss": 0.1789,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5506451285177283,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.1397161821428973e-05,
+      "loss": 0.2187,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5516610789393478,
+      "grad_norm": 2.25,
+      "learning_rate": 2.131741064735212e-05,
+      "loss": 0.1367,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5526770293609672,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.1237697761678684e-05,
+      "loss": 0.1574,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5536929797825866,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.1158023993194848e-05,
+      "loss": 0.1301,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5547089302042061,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.107839017028005e-05,
+      "loss": 0.2782,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5557248806258255,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.0998797120898457e-05,
+      "loss": 0.2024,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5567408310474449,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0919245672590277e-05,
+      "loss": 0.1755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5577567814690643,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.083973665246318e-05,
+      "loss": 0.2058,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5587727318906838,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.076027088718373e-05,
+      "loss": 0.2159,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5597886823123032,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.0680849202968743e-05,
+      "loss": 0.2139,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5608046327339226,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.060147242557674e-05,
+      "loss": 0.183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5618205831555421,
+      "grad_norm": 5.5,
+      "learning_rate": 2.0522141380299308e-05,
+      "loss": 0.1673,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5628365335771615,
+      "grad_norm": 4.25,
+      "learning_rate": 2.044285689195258e-05,
+      "loss": 0.1674,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5638524839987809,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.0363619784868604e-05,
+      "loss": 0.1531,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5648684344204002,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0284430882886836e-05,
+      "loss": 0.1665,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5658843848420197,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.020529100934549e-05,
+      "loss": 0.1717,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5669003352636391,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.012620098707306e-05,
+      "loss": 0.1167,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5679162856852585,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.004716163837972e-05,
+      "loss": 0.2084,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5689322361068779,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.996817378504876e-05,
+      "loss": 0.1939,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5699481865284974,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.9889238248328108e-05,
+      "loss": 0.1241,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5709641369501168,
+      "grad_norm": 4.875,
+      "learning_rate": 1.981035584892171e-05,
+      "loss": 0.1865,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5719800873717362,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9731527406981072e-05,
+      "loss": 0.1639,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5729960377933557,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.9652753742096655e-05,
+      "loss": 0.2019,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5740119882149751,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.9574035673289432e-05,
+      "loss": 0.1829,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5750279386365945,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9495374019002312e-05,
+      "loss": 0.2267,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5760438890582139,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9416769597091673e-05,
+      "loss": 0.1411,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5770598394798334,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.9338223224818818e-05,
+      "loss": 0.1476,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5780757899014528,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9259735718841524e-05,
+      "loss": 0.1417,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5790917403230722,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.918130789520551e-05,
+      "loss": 0.1592,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5801076907446917,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9102940569335963e-05,
+      "loss": 0.161,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5811236411663111,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.9024634556029093e-05,
+      "loss": 0.1614,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5821395915879305,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.89463906694436e-05,
+      "loss": 0.1505,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5831555420095499,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8868209723092286e-05,
+      "loss": 0.1674,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5841714924311694,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.8790092529833508e-05,
+      "loss": 0.1468,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5851874428527888,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.871203990186281e-05,
+      "loss": 0.1903,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5862033932744082,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8634052650704415e-05,
+      "loss": 0.2644,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5872193436960277,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8556131587202848e-05,
+      "loss": 0.1968,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.8478277521514424e-05,
+      "loss": 0.2249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5892512445392665,
+      "grad_norm": 4.0,
+      "learning_rate": 1.8400491263098906e-05,
+      "loss": 0.1881,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5902671949608859,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.832277362071106e-05,
+      "loss": 0.1352,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5912831453825054,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.824512540239221e-05,
+      "loss": 0.2737,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5922990958041248,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.81675474154619e-05,
+      "loss": 0.1566,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5933150462257442,
+      "grad_norm": 2.6875,
+      "learning_rate": 1.8090040466509444e-05,
+      "loss": 0.1999,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5943309966473637,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.8012605361385592e-05,
+      "loss": 0.2372,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5953469470689831,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7935242905194087e-05,
+      "loss": 0.2411,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5963628974906025,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.785795390228336e-05,
+      "loss": 0.138,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5973788479122218,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.7780739156238125e-05,
+      "loss": 0.1867,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5983947983338413,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.770359946987105e-05,
+      "loss": 0.2091,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5994107487554607,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.7626535645214378e-05,
+      "loss": 0.2091,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6004266991770801,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.7549548483511614e-05,
+      "loss": 0.1927,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6014426495986995,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.7472638785209198e-05,
+      "loss": 0.1893,
+      "step": 5920
+    },
+    {
+      "epoch": 0.602458600020319,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.7395807349948145e-05,
+      "loss": 0.1557,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6034745504419384,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.73190549765558e-05,
+      "loss": 0.1717,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6044905008635578,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.724238246303745e-05,
+      "loss": 0.1879,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6055064512851773,
+      "grad_norm": 3.875,
+      "learning_rate": 1.71657906065681e-05,
+      "loss": 0.1908,
+      "step": 5960
+    },
+    {
+      "epoch": 0.6065224017067967,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.7089280203484115e-05,
+      "loss": 0.1712,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6075383521284161,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.701285204927502e-05,
+      "loss": 0.1454,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6085543025500355,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.693650693857515e-05,
+      "loss": 0.2283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.609570252971655,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6860245665155466e-05,
+      "loss": 0.2188,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6105862033932744,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.678406902191521e-05,
+      "loss": 0.1605,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6116021538148938,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.670797780087374e-05,
+      "loss": 0.1472,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6126181042365133,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.6631972793162288e-05,
+      "loss": 0.1676,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6136340546581327,
+      "grad_norm": 1.25,
+      "learning_rate": 1.6556054789015662e-05,
+      "loss": 0.1508,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6146500050797521,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.6480224577764132e-05,
+      "loss": 0.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6156659555013715,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6404482947825137e-05,
+      "loss": 0.2514,
+      "step": 6060
+    },
+    {
+      "epoch": 0.616681905922991,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.6328830686695154e-05,
+      "loss": 0.2397,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6176978563446104,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.625326858094144e-05,
+      "loss": 0.1523,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6187138067662298,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6177797416193953e-05,
+      "loss": 0.218,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6197297571878493,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.6102417977137052e-05,
+      "loss": 0.1476,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6207457076094687,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.602713104750147e-05,
+      "loss": 0.1818,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 4.375,
+      "learning_rate": 1.5951937410056087e-05,
+      "loss": 0.2061,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6227776084527075,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.587683784659979e-05,
+      "loss": 0.1566,
+      "step": 6130
+    },
+    {
+      "epoch": 0.623793558874327,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.58018331379534e-05,
+      "loss": 0.1376,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6248095092959464,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.572692406395149e-05,
+      "loss": 0.1655,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6258254597175658,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.5652111403434338e-05,
+      "loss": 0.2363,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6268414101391853,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.5577395934239757e-05,
+      "loss": 0.2464,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6278573605608047,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.5502778433195085e-05,
+      "loss": 0.1898,
+      "step": 6180
+    },
+    {
+      "epoch": 0.628873310982424,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5428259676109048e-05,
+      "loss": 0.1804,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6298892614040434,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.5353840437763732e-05,
+      "loss": 0.1409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.630905211825663,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5279521491906496e-05,
+      "loss": 0.2449,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6319211622472823,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.520530361124195e-05,
+      "loss": 0.2103,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6329371126689017,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.5131187567423937e-05,
+      "loss": 0.2156,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6339530630905211,
+      "grad_norm": 2.703125,
+      "learning_rate": 1.5057174131047446e-05,
+      "loss": 0.161,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6349690135121406,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4983264071640679e-05,
+      "loss": 0.1757,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63598496393376,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.490945815765699e-05,
+      "loss": 0.2011,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6370009143553794,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4835757156466945e-05,
+      "loss": 0.1658,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6380168647769989,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.4762161834350271e-05,
+      "loss": 0.1754,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6390328151986183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4688672956487987e-05,
+      "loss": 0.1427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6400487656202377,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.4615291286954352e-05,
+      "loss": 0.1517,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6410647160418571,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4542017588709005e-05,
+      "loss": 0.2348,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6420806664634766,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.4468852623588961e-05,
+      "loss": 0.2089,
+      "step": 6320
+    },
+    {
+      "epoch": 0.643096616885096,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4395797152300719e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6441125673067154,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4322851934412382e-05,
+      "loss": 0.1017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6451285177283349,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.4250017728345716e-05,
+      "loss": 0.1813,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6461444681499543,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4177295291368292e-05,
+      "loss": 0.1095,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6471604185715737,
+      "grad_norm": 2.625,
+      "learning_rate": 1.410468537958558e-05,
+      "loss": 0.2259,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6481763689931931,
+      "grad_norm": 3.5,
+      "learning_rate": 1.4032188747933136e-05,
+      "loss": 0.1595,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6491923194148126,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.39598061501687e-05,
+      "loss": 0.2226,
+      "step": 6390
+    },
+    {
+      "epoch": 0.650208269836432,
+      "grad_norm": 5.34375,
+      "learning_rate": 1.388753833886442e-05,
+      "loss": 0.2132,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6512242202580514,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3815386065398945e-05,
+      "loss": 0.1227,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6522401706796709,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3743350079949705e-05,
+      "loss": 0.1755,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6532561211012903,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.3671431131485057e-05,
+      "loss": 0.1552,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6542720715229097,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.3599629967756483e-05,
+      "loss": 0.1917,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6552880219445291,
+      "grad_norm": 4.625,
+      "learning_rate": 1.3527947335290877e-05,
+      "loss": 0.1812,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6563039723661486,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.3456383979382708e-05,
+      "loss": 0.1896,
+      "step": 6460
+    },
+    {
+      "epoch": 0.657319922787768,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.3384940644086352e-05,
+      "loss": 0.1484,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6583358732093874,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.3313618072208268e-05,
+      "loss": 0.1334,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6593518236310069,
+      "grad_norm": 4.375,
+      "learning_rate": 1.3242417005299357e-05,
+      "loss": 0.1351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6603677740526263,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.31713381836472e-05,
+      "loss": 0.1717,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6613837244742456,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.3100382346268392e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 0.662399674895865,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.3029550230900812e-05,
+      "loss": 0.1997,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6634156253174845,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2958842573996016e-05,
+      "loss": 0.1969,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6644315757391039,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2888260110711525e-05,
+      "loss": 0.1469,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6654475261607233,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.2817803574903212e-05,
+      "loss": 0.1524,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6664634765823427,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.2747473699117668e-05,
+      "loss": 0.159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6674794270039622,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.267727121458458e-05,
+      "loss": 0.1999,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6684953774255816,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.2607196851209137e-05,
+      "loss": 0.2216,
+      "step": 6580
+    },
+    {
+      "epoch": 0.669511327847201,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2537251337564412e-05,
+      "loss": 0.1607,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6705272782688205,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.2467435400883839e-05,
+      "loss": 0.2187,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6715432286904399,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.239774976705359e-05,
+      "loss": 0.1753,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6725591791120593,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2328195160605092e-05,
+      "loss": 0.194,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6735751295336787,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.225877230470743e-05,
+      "loss": 0.1485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6745910799552982,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.218948192115988e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6756070303769176,
+      "grad_norm": 3.875,
+      "learning_rate": 1.21203247303844e-05,
+      "loss": 0.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.676622980798537,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.2051301451418073e-05,
+      "loss": 0.2377,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6776389312201565,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.198241280190574e-05,
+      "loss": 0.1508,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6786548816417759,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.1913659498092431e-05,
+      "loss": 0.1537,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6796708320633953,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.184504225481601e-05,
+      "loss": 0.2339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6806867824850147,
+      "grad_norm": 5.625,
+      "learning_rate": 1.177656178549966e-05,
+      "loss": 0.2102,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6817027329066342,
+      "grad_norm": 2.5,
+      "learning_rate": 1.1708218802144536e-05,
+      "loss": 0.1435,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6827186833282536,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1640014015322323e-05,
+      "loss": 0.1823,
+      "step": 6720
+    },
+    {
+      "epoch": 0.683734633749873,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1571948134167862e-05,
+      "loss": 0.1154,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6847505841714925,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1504021866371761e-05,
+      "loss": 0.2105,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6857665345931119,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.143623591817304e-05,
+      "loss": 0.1317,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6867824850147313,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.1368590994351835e-05,
+      "loss": 0.1406,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6877984354363507,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.130108779822198e-05,
+      "loss": 0.1425,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6888143858579702,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.1233727031623783e-05,
+      "loss": 0.1623,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6898303362795896,
+      "grad_norm": 4.625,
+      "learning_rate": 1.1166509394916682e-05,
+      "loss": 0.1591,
+      "step": 6790
+    },
+    {
+      "epoch": 0.690846286701209,
+      "grad_norm": 3.84375,
+      "learning_rate": 1.1099435586971982e-05,
+      "loss": 0.1758,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6918622371228285,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.1032506305165555e-05,
+      "loss": 0.1018,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6928781875444479,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.0965722245370641e-05,
+      "loss": 0.1485,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6938941379660672,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0899084101950561e-05,
+      "loss": 0.1762,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6949100883876866,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.0832592567751555e-05,
+      "loss": 0.1402,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6959260388093061,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0766248334095505e-05,
+      "loss": 0.2278,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6969419892309255,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.0700052090772828e-05,
+      "loss": 0.1969,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6979579396525449,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.0634004526035249e-05,
+      "loss": 0.2073,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6989738900741643,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.0568106326588645e-05,
+      "loss": 0.1902,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6999898404957838,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.0502358177585953e-05,
+      "loss": 0.2165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7010057909174032,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.0436760762619977e-05,
+      "loss": 0.1952,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7020217413390226,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.0371314763716347e-05,
+      "loss": 0.1422,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7030376917606421,
+      "grad_norm": 2.53125,
+      "learning_rate": 1.0306020861326388e-05,
+      "loss": 0.0961,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7040536421822615,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.0240879734320068e-05,
+      "loss": 0.1542,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7050695926038809,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.0175892059978901e-05,
+      "loss": 0.1748,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7060855430255003,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.0111058513988958e-05,
+      "loss": 0.0819,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7071014934471198,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.0046379770433803e-05,
+      "loss": 0.1933,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7081174438687392,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.98185650178749e-06,
+      "loss": 0.1891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7091333942903586,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.917489378907591e-06,
+      "loss": 0.2102,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7101493447119781,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.853279071028212e-06,
+      "loss": 0.1714,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7111652951335975,
+      "grad_norm": 2.375,
+      "learning_rate": 9.78922624575303e-06,
+      "loss": 0.1299,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7121812455552169,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.72533156904833e-06,
+      "loss": 0.1914,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7131971959768363,
+      "grad_norm": 3.859375,
+      "learning_rate": 9.661595705236137e-06,
+      "loss": 0.2377,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7142131463984558,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.598019316987244e-06,
+      "loss": 0.1851,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7152290968200752,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.53460306531439e-06,
+      "loss": 0.2661,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7162450472416946,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.471347609565311e-06,
+      "loss": 0.1669,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7172609976633141,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.408253607415957e-06,
+      "loss": 0.2487,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7182769480849335,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.345321714863614e-06,
+      "loss": 0.186,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7192928985065529,
+      "grad_norm": 6.0625,
+      "learning_rate": 9.282552586220075e-06,
+      "loss": 0.2249,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7203088489281723,
+      "grad_norm": 1.5703125,
+      "learning_rate": 9.219946874104885e-06,
+      "loss": 0.1255,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7213247993497918,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.157505229438481e-06,
+      "loss": 0.1999,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7223407497714112,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.095228301435518e-06,
+      "loss": 0.199,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7233567001930306,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.03311673759802e-06,
+      "loss": 0.2182,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7243726506146501,
+      "grad_norm": 6.46875,
+      "learning_rate": 8.971171183708733e-06,
+      "loss": 0.1573,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7253886010362695,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.909392283824353e-06,
+      "loss": 0.2044,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7264045514578888,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.847780680268872e-06,
+      "loss": 0.11,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7274205018795082,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.786337013626853e-06,
+      "loss": 0.1897,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7284364523011277,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.725061922736799e-06,
+      "loss": 0.153,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7294524027227471,
+      "grad_norm": 1.609375,
+      "learning_rate": 8.663956044684532e-06,
+      "loss": 0.1746,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7304683531443665,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.603020014796507e-06,
+      "loss": 0.2284,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7314843035659859,
+      "grad_norm": 1.515625,
+      "learning_rate": 8.542254466633273e-06,
+      "loss": 0.1186,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7325002539876054,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.481660031982844e-06,
+      "loss": 0.1971,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7335162044092248,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.421237340854157e-06,
+      "loss": 0.196,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7345321548308442,
+      "grad_norm": 0.65234375,
+      "learning_rate": 8.360987021470479e-06,
+      "loss": 0.1724,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7355481052524637,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.300909700262929e-06,
+      "loss": 0.175,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7365640556740831,
+      "grad_norm": 3.109375,
+      "learning_rate": 8.241006001863924e-06,
+      "loss": 0.2276,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7375800060957025,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.181276549100714e-06,
+      "loss": 0.2029,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7385959565173219,
+      "grad_norm": 4.03125,
+      "learning_rate": 8.12172196298887e-06,
+      "loss": 0.175,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7396119069389414,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.062342862725878e-06,
+      "loss": 0.1662,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7406278573605608,
+      "grad_norm": 3.375,
+      "learning_rate": 8.003139865684662e-06,
+      "loss": 0.1616,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7416438077821802,
+      "grad_norm": 2.5625,
+      "learning_rate": 7.944113587407157e-06,
+      "loss": 0.2448,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7426597582037997,
+      "grad_norm": 4.125,
+      "learning_rate": 7.885264641597961e-06,
+      "loss": 0.1618,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7436757086254191,
+      "grad_norm": 3.5,
+      "learning_rate": 7.826593640117889e-06,
+      "loss": 0.1134,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7446916590470385,
+      "grad_norm": 2.6875,
+      "learning_rate": 7.76810119297767e-06,
+      "loss": 0.1795,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7457076094686579,
+      "grad_norm": 4.34375,
+      "learning_rate": 7.709787908331556e-06,
+      "loss": 0.2736,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7467235598902774,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.651654392471038e-06,
+      "loss": 0.139,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7477395103118968,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.593701249818521e-06,
+      "loss": 0.2023,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7487554607335162,
+      "grad_norm": 2.15625,
+      "learning_rate": 7.535929082921048e-06,
+      "loss": 0.1702,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7497714111551357,
+      "grad_norm": 1.96875,
+      "learning_rate": 7.47833849244402e-06,
+      "loss": 0.1835,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7507873615767551,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.420930077164959e-06,
+      "loss": 0.1713,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7518033119983745,
+      "grad_norm": 4.46875,
+      "learning_rate": 7.363704433967311e-06,
+      "loss": 0.1906,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7528192624199939,
+      "grad_norm": 1.75,
+      "learning_rate": 7.306662157834185e-06,
+      "loss": 0.1421,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7538352128416134,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.2498038418422145e-06,
+      "loss": 0.1793,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7548511632632328,
+      "grad_norm": 2.578125,
+      "learning_rate": 7.193130077155374e-06,
+      "loss": 0.1603,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7558671136848522,
+      "grad_norm": 4.3125,
+      "learning_rate": 7.13664145301883e-06,
+      "loss": 0.2169,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7568830641064717,
+      "grad_norm": 3.078125,
+      "learning_rate": 7.0803385567528025e-06,
+      "loss": 0.1685,
+      "step": 7450
+    },
+    {
+      "epoch": 0.757899014528091,
+      "grad_norm": 3.5625,
+      "learning_rate": 7.024221973746495e-06,
+      "loss": 0.2282,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7589149649497104,
+      "grad_norm": 2.265625,
+      "learning_rate": 6.968292287451961e-06,
+      "loss": 0.1786,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7599309153713298,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.912550079378091e-06,
+      "loss": 0.1811,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7609468657929493,
+      "grad_norm": 2.328125,
+      "learning_rate": 6.856995929084506e-06,
+      "loss": 0.1747,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7619628162145687,
+      "grad_norm": 5.21875,
+      "learning_rate": 6.801630414175589e-06,
+      "loss": 0.2028,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7629787666361881,
+      "grad_norm": 3.78125,
+      "learning_rate": 6.746454110294451e-06,
+      "loss": 0.2255,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7639947170578075,
+      "grad_norm": 1.625,
+      "learning_rate": 6.691467591116931e-06,
+      "loss": 0.1604,
+      "step": 7520
+    },
+    {
+      "epoch": 0.765010667479427,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.6366714283456755e-06,
+      "loss": 0.2559,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7660266179010464,
+      "grad_norm": 4.59375,
+      "learning_rate": 6.582066191704142e-06,
+      "loss": 0.2034,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7670425683226658,
+      "grad_norm": 1.578125,
+      "learning_rate": 6.527652448930724e-06,
+      "loss": 0.148,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7680585187442853,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.4734307657728e-06,
+      "loss": 0.1811,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7690744691659047,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.419401705980924e-06,
+      "loss": 0.1407,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7700904195875241,
+      "grad_norm": 2.25,
+      "learning_rate": 6.365565831302869e-06,
+      "loss": 0.1893,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7711063700091435,
+      "grad_norm": 1.625,
+      "learning_rate": 6.311923701477854e-06,
+      "loss": 0.1835,
+      "step": 7590
+    },
+    {
+      "epoch": 0.772122320430763,
+      "grad_norm": 2.375,
+      "learning_rate": 6.258475874230713e-06,
+      "loss": 0.1579,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7731382708523824,
+      "grad_norm": 4.5,
+      "learning_rate": 6.205222905266067e-06,
+      "loss": 0.1794,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7741542212740018,
+      "grad_norm": 4.25,
+      "learning_rate": 6.152165348262598e-06,
+      "loss": 0.1477,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7751701716956213,
+      "grad_norm": 1.9765625,
+      "learning_rate": 6.0993037548672246e-06,
+      "loss": 0.2396,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7761861221172407,
+      "grad_norm": 2.671875,
+      "learning_rate": 6.046638674689454e-06,
+      "loss": 0.1717,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7772020725388601,
+      "grad_norm": 3.671875,
+      "learning_rate": 5.994170655295567e-06,
+      "loss": 0.2646,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7782180229604795,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.9419002422030106e-06,
+      "loss": 0.1553,
+      "step": 7660
+    },
+    {
+      "epoch": 0.779233973382099,
+      "grad_norm": 3.734375,
+      "learning_rate": 5.889827978874665e-06,
+      "loss": 0.1854,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7802499238037184,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.837954406713245e-06,
+      "loss": 0.1857,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7812658742253378,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.786280065055619e-06,
+      "loss": 0.1797,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7822818246469573,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.734805491167244e-06,
+      "loss": 0.1488,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7832977750685767,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.683531220236576e-06,
+      "loss": 0.1688,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.632457785369455e-06,
+      "loss": 0.1503,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7853296759118155,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.581585717583637e-06,
+      "loss": 0.1658,
+      "step": 7730
+    },
+    {
+      "epoch": 0.786345626333435,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.530915545803209e-06,
+      "loss": 0.2112,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7873615767550544,
+      "grad_norm": 4.1875,
+      "learning_rate": 5.480447796853141e-06,
+      "loss": 0.165,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7883775271766738,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.430182995453756e-06,
+      "loss": 0.1499,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7893934775982933,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.380121664215329e-06,
+      "loss": 0.1559,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7904094280199127,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.330264323632611e-06,
+      "loss": 0.2098,
+      "step": 7780
+    },
+    {
+      "epoch": 0.791425378441532,
+      "grad_norm": 4.65625,
+      "learning_rate": 5.280611492079449e-06,
+      "loss": 0.1776,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7924413288631514,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.231163685803361e-06,
+      "loss": 0.1497,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7934572792847709,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.181921418920191e-06,
+      "loss": 0.12,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7944732297063903,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.13288520340878e-06,
+      "loss": 0.1981,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7954891801280097,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.084055549105596e-06,
+      "loss": 0.1389,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7965051305496291,
+      "grad_norm": 2.796875,
+      "learning_rate": 5.035432963699479e-06,
+      "loss": 0.2293,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7975210809712486,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.98701795272635e-06,
+      "loss": 0.1618,
+      "step": 7850
+    },
+    {
+      "epoch": 0.798537031392868,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.938811019563938e-06,
+      "loss": 0.1755,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7995529818144874,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.8908126654265475e-06,
+      "loss": 0.1565,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8005689322361069,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.843023389359885e-06,
+      "loss": 0.2176,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8015848826577263,
+      "grad_norm": 2.625,
+      "learning_rate": 4.79544368823581e-06,
+      "loss": 0.2013,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8026008330793457,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.748074056747234e-06,
+      "loss": 0.1246,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8036167835009651,
+      "grad_norm": 3.5,
+      "learning_rate": 4.700914987402919e-06,
+      "loss": 0.1638,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8046327339225846,
+      "grad_norm": 3.4375,
+      "learning_rate": 4.6539669705223916e-06,
+      "loss": 0.2213,
+      "step": 7920
+    },
+    {
+      "epoch": 0.805648684344204,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.607230494230849e-06,
+      "loss": 0.1822,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8066646347658234,
+      "grad_norm": 2.359375,
+      "learning_rate": 4.560706044454047e-06,
+      "loss": 0.1763,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8076805851874429,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.514394104913291e-06,
+      "loss": 0.234,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8086965356090623,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.468295157120372e-06,
+      "loss": 0.1939,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8097124860306817,
+      "grad_norm": 2.578125,
+      "learning_rate": 4.422409680372594e-06,
+      "loss": 0.174,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8107284364523011,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.3767381517477505e-06,
+      "loss": 0.2375,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8117443868739206,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.331281046099203e-06,
+      "loss": 0.2076,
+      "step": 7990
+    },
+    {
+      "epoch": 0.81276033729554,
+      "grad_norm": 6.0625,
+      "learning_rate": 4.286038836050929e-06,
+      "loss": 0.2504,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8137762877171594,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.241011991992586e-06,
+      "loss": 0.2102,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8147922381387789,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.1962009820746635e-06,
+      "loss": 0.1846,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8158081885603983,
+      "grad_norm": 1.875,
+      "learning_rate": 4.15160627220357e-06,
+      "loss": 0.1741,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8168241389820177,
+      "grad_norm": 5.5625,
+      "learning_rate": 4.107228326036838e-06,
+      "loss": 0.2078,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8178400894036371,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.063067604978252e-06,
+      "loss": 0.212,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8188560398252566,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.019124568173094e-06,
+      "loss": 0.1831,
+      "step": 8060
+    },
+    {
+      "epoch": 0.819871990246876,
+      "grad_norm": 6.625,
+      "learning_rate": 3.975399672503341e-06,
+      "loss": 0.2196,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8208879406684954,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.931893372582943e-06,
+      "loss": 0.2002,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8219038910901149,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.888606120753047e-06,
+      "loss": 0.2138,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8229198415117343,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.845538367077362e-06,
+      "loss": 0.2593,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8239357919333536,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.8026905593374213e-06,
+      "loss": 0.2062,
+      "step": 8110
+    },
+    {
+      "epoch": 0.824951742354973,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.760063143027945e-06,
+      "loss": 0.1343,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8259676927765925,
+      "grad_norm": 1.984375,
+      "learning_rate": 3.7176565613522313e-06,
+      "loss": 0.2494,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8269836431982119,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.675471255217516e-06,
+      "loss": 0.1502,
+      "step": 8140
+    },
+    {
+      "epoch": 0.8279995936198313,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.6335076632304175e-06,
+      "loss": 0.1256,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8290155440414507,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.5917662216923332e-06,
+      "loss": 0.1709,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8300314944630702,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.550247364594958e-06,
+      "loss": 0.1881,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8310474448846896,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.508951523615725e-06,
+      "loss": 0.1998,
+      "step": 8180
+    },
+    {
+      "epoch": 0.832063395306309,
+      "grad_norm": 2.40625,
+      "learning_rate": 3.467879128113352e-06,
+      "loss": 0.2429,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8330793457279285,
+      "grad_norm": 2.609375,
+      "learning_rate": 3.427030605123352e-06,
+      "loss": 0.1942,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8340952961495479,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.3864063793536043e-06,
+      "loss": 0.1898,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8351112465711673,
+      "grad_norm": 5.375,
+      "learning_rate": 3.3460068731799577e-06,
+      "loss": 0.1919,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8361271969927867,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.3058325066417818e-06,
+      "loss": 0.1516,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8371431474144062,
+      "grad_norm": 0.76171875,
+      "learning_rate": 3.26588369743768e-06,
+      "loss": 0.1068,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8381590978360256,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.2261608609210653e-06,
+      "loss": 0.1203,
+      "step": 8250
+    },
+    {
+      "epoch": 0.839175048257645,
+      "grad_norm": 2.359375,
+      "learning_rate": 3.186664410095913e-06,
+      "loss": 0.2172,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8401909986792645,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.1473947556124093e-06,
+      "loss": 0.1249,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8412069491008839,
+      "grad_norm": 2.484375,
+      "learning_rate": 3.1083523057627213e-06,
+      "loss": 0.1744,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8422228995225033,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.0695374664767353e-06,
+      "loss": 0.1772,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8432388499441227,
+      "grad_norm": 0.59375,
+      "learning_rate": 3.0309506413178397e-06,
+      "loss": 0.2302,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8442548003657422,
+      "grad_norm": 2.390625,
+      "learning_rate": 2.9925922314787136e-06,
+      "loss": 0.1635,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8452707507873616,
+      "grad_norm": 2.34375,
+      "learning_rate": 2.954462635777194e-06,
+      "loss": 0.1573,
+      "step": 8320
+    },
+    {
+      "epoch": 0.846286701208981,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.916562250652083e-06,
+      "loss": 0.1608,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8473026516306005,
+      "grad_norm": 4.125,
+      "learning_rate": 2.878891470159048e-06,
+      "loss": 0.184,
+      "step": 8340
+    },
+    {
+      "epoch": 0.8483186020522199,
+      "grad_norm": 2.515625,
+      "learning_rate": 2.8414506859665514e-06,
+      "loss": 0.2141,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8493345524738393,
+      "grad_norm": 3.375,
+      "learning_rate": 2.8042402873517197e-06,
+      "loss": 0.1729,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8503505028954587,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.76726066119635e-06,
+      "loss": 0.2252,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8513664533170782,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.730512191982845e-06,
+      "loss": 0.1644,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8523824037386976,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.693995261790261e-06,
+      "loss": 0.1822,
+      "step": 8390
+    },
+    {
+      "epoch": 0.853398354160317,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.657710250290285e-06,
+      "loss": 0.2068,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8544143045819365,
+      "grad_norm": 0.640625,
+      "learning_rate": 2.621657534743327e-06,
+      "loss": 0.1224,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8554302550035559,
+      "grad_norm": 3.421875,
+      "learning_rate": 2.5858374899945804e-06,
+      "loss": 0.179,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8564462054251752,
+      "grad_norm": 3.484375,
+      "learning_rate": 2.550250488470135e-06,
+      "loss": 0.1873,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8574621558467946,
+      "grad_norm": 3.984375,
+      "learning_rate": 2.5148969001730806e-06,
+      "loss": 0.1799,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8584781062684141,
+      "grad_norm": 1.375,
+      "learning_rate": 2.4797770926796858e-06,
+      "loss": 0.176,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8594940566900335,
+      "grad_norm": 1.8984375,
+      "learning_rate": 2.444891431135571e-06,
+      "loss": 0.1664,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8605100071116529,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.4102402782518936e-06,
+      "loss": 0.1512,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8615259575332723,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.3758239943016096e-06,
+      "loss": 0.1629,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8625419079548918,
+      "grad_norm": 5.3125,
+      "learning_rate": 2.3416429371157013e-06,
+      "loss": 0.2099,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8635578583765112,
+      "grad_norm": 5.9375,
+      "learning_rate": 2.307697462079464e-06,
+      "loss": 0.2221,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8645738087981306,
+      "grad_norm": 5.4375,
+      "learning_rate": 2.273987922128809e-06,
+      "loss": 0.2191,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8655897592197501,
+      "grad_norm": 2.171875,
+      "learning_rate": 2.240514667746607e-06,
+      "loss": 0.1843,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8666057096413695,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.2072780469590245e-06,
+      "loss": 0.2494,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8676216600629889,
+      "grad_norm": 2.25,
+      "learning_rate": 2.1742784053319116e-06,
+      "loss": 0.1712,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8686376104846083,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.141516085967224e-06,
+      "loss": 0.1169,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8696535609062278,
+      "grad_norm": 4.25,
+      "learning_rate": 2.1089914294994434e-06,
+      "loss": 0.1374,
+      "step": 8560
+    },
+    {
+      "epoch": 0.8706695113278472,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.0767047740920336e-06,
+      "loss": 0.2162,
+      "step": 8570
+    },
+    {
+      "epoch": 0.8716854617494666,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.0446564554339187e-06,
+      "loss": 0.1593,
+      "step": 8580
+    },
+    {
+      "epoch": 0.8727014121710861,
+      "grad_norm": 2.671875,
+      "learning_rate": 2.0128468067360185e-06,
+      "loss": 0.1857,
+      "step": 8590
+    },
+    {
+      "epoch": 0.8737173625927055,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.981276158727749e-06,
+      "loss": 0.1989,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8747333130143249,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.949944839653625e-06,
+      "loss": 0.2077,
+      "step": 8610
+    },
+    {
+      "epoch": 0.8757492634359443,
+      "grad_norm": 2.625,
+      "learning_rate": 1.918853175269797e-06,
+      "loss": 0.2003,
+      "step": 8620
+    },
+    {
+      "epoch": 0.8767652138575638,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.8880014888407127e-06,
+      "loss": 0.2486,
+      "step": 8630
+    },
+    {
+      "epoch": 0.8777811642791832,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.8573901011357336e-06,
+      "loss": 0.1896,
+      "step": 8640
+    },
+    {
+      "epoch": 0.8787971147008026,
+      "grad_norm": 5.0625,
+      "learning_rate": 1.8270193304257887e-06,
+      "loss": 0.1727,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8798130651224221,
+      "grad_norm": 1.75,
+      "learning_rate": 1.7968894924800916e-06,
+      "loss": 0.1687,
+      "step": 8660
+    },
+    {
+      "epoch": 0.8808290155440415,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.7670009005628291e-06,
+      "loss": 0.166,
+      "step": 8670
+    },
+    {
+      "epoch": 0.8818449659656609,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.737353865429936e-06,
+      "loss": 0.1471,
+      "step": 8680
+    },
+    {
+      "epoch": 0.8828609163872803,
+      "grad_norm": 0.546875,
+      "learning_rate": 1.7079486953258283e-06,
+      "loss": 0.1075,
+      "step": 8690
+    },
+    {
+      "epoch": 0.8838768668088998,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.6787856959802367e-06,
+      "loss": 0.2113,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8848928172305192,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.6498651706049945e-06,
+      "loss": 0.1412,
+      "step": 8710
+    },
+    {
+      "epoch": 0.8859087676521386,
+      "grad_norm": 3.796875,
+      "learning_rate": 1.6211874198909072e-06,
+      "loss": 0.1701,
+      "step": 8720
+    },
+    {
+      "epoch": 0.8869247180737581,
+      "grad_norm": 3.734375,
+      "learning_rate": 1.592752742004605e-06,
+      "loss": 0.1348,
+      "step": 8730
+    },
+    {
+      "epoch": 0.8879406684953774,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.5645614325854735e-06,
+      "loss": 0.1931,
+      "step": 8740
+    },
+    {
+      "epoch": 0.8889566189169968,
+      "grad_norm": 3.4375,
+      "learning_rate": 1.5366137847425466e-06,
+      "loss": 0.1705,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8899725693386162,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.5089100890514769e-06,
+      "loss": 0.1889,
+      "step": 8760
+    },
+    {
+      "epoch": 0.8909885197602357,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.4814506335515176e-06,
+      "loss": 0.1837,
+      "step": 8770
+    },
+    {
+      "epoch": 0.8920044701818551,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.4542357037425207e-06,
+      "loss": 0.1728,
+      "step": 8780
+    },
+    {
+      "epoch": 0.8930204206034745,
+      "grad_norm": 1.625,
+      "learning_rate": 1.4272655825819713e-06,
+      "loss": 0.1562,
+      "step": 8790
+    },
+    {
+      "epoch": 0.8940363710250939,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.4005405504820351e-06,
+      "loss": 0.1681,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8950523214467134,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.3740608853066634e-06,
+      "loss": 0.1449,
+      "step": 8810
+    },
+    {
+      "epoch": 0.8960682718683328,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.347826862368684e-06,
+      "loss": 0.2418,
+      "step": 8820
+    },
+    {
+      "epoch": 0.8970842222899522,
+      "grad_norm": 0.55859375,
+      "learning_rate": 1.3218387544269545e-06,
+      "loss": 0.2473,
+      "step": 8830
+    },
+    {
+      "epoch": 0.8981001727115717,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.2960968316835132e-06,
+      "loss": 0.194,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8991161231331911,
+      "grad_norm": 3.921875,
+      "learning_rate": 1.2706013617807822e-06,
+      "loss": 0.2109,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9001320735548105,
+      "grad_norm": 5.03125,
+      "learning_rate": 1.2453526097987778e-06,
+      "loss": 0.151,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9011480239764299,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.2203508382523431e-06,
+      "loss": 0.1811,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9021639743980494,
+      "grad_norm": 3.828125,
+      "learning_rate": 1.1955963070884534e-06,
+      "loss": 0.2004,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9031799248196688,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.171089273683465e-06,
+      "loss": 0.1395,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9041958752412882,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.1468299928404868e-06,
+      "loss": 0.1915,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9052118256629077,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.1228187167866943e-06,
+      "loss": 0.1281,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9062277760845271,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.099055695170728e-06,
+      "loss": 0.1627,
+      "step": 8920
+    },
+    {
+      "epoch": 0.9072437265061465,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.0755411750600962e-06,
+      "loss": 0.1768,
+      "step": 8930
+    },
+    {
+      "epoch": 0.9082596769277659,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.052275400938596e-06,
+      "loss": 0.1544,
+      "step": 8940
+    },
+    {
+      "epoch": 0.9092756273493854,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.0292586147037764e-06,
+      "loss": 0.2498,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9102915777710048,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.0064910556644214e-06,
+      "loss": 0.1918,
+      "step": 8960
+    },
+    {
+      "epoch": 0.9113075281926242,
+      "grad_norm": 4.0,
+      "learning_rate": 9.839729605380766e-07,
+      "loss": 0.2388,
+      "step": 8970
+    },
+    {
+      "epoch": 0.9123234786142437,
+      "grad_norm": 3.765625,
+      "learning_rate": 9.61704563448565e-07,
+      "loss": 0.1944,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9133394290358631,
+      "grad_norm": 2.90625,
+      "learning_rate": 9.396860959235671e-07,
+      "loss": 0.1667,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9143553794574825,
+      "grad_norm": 2.4375,
+      "learning_rate": 9.179177868922085e-07,
+      "loss": 0.2143,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9153713298791019,
+      "grad_norm": 3.03125,
+      "learning_rate": 8.963998626826925e-07,
+      "loss": 0.1994,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9163872803007214,
+      "grad_norm": 3.859375,
+      "learning_rate": 8.751325470199134e-07,
+      "loss": 0.1714,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9174032307223408,
+      "grad_norm": 3.375,
+      "learning_rate": 8.541160610231803e-07,
+      "loss": 0.144,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9184191811439602,
+      "grad_norm": 1.3046875,
+      "learning_rate": 8.333506232038629e-07,
+      "loss": 0.1333,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9194351315655795,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.128364494631724e-07,
+      "loss": 0.1504,
+      "step": 9050
+    },
+    {
+      "epoch": 0.920451081987199,
+      "grad_norm": 2.90625,
+      "learning_rate": 7.925737530898702e-07,
+      "loss": 0.2235,
+      "step": 9060
+    },
+    {
+      "epoch": 0.9214670324088184,
+      "grad_norm": 3.359375,
+      "learning_rate": 7.725627447580902e-07,
+      "loss": 0.1256,
+      "step": 9070
+    },
+    {
+      "epoch": 0.9224829828304378,
+      "grad_norm": 6.125,
+      "learning_rate": 7.528036325251231e-07,
+      "loss": 0.1963,
+      "step": 9080
+    },
+    {
+      "epoch": 0.9234989332520573,
+      "grad_norm": 2.359375,
+      "learning_rate": 7.33296621829252e-07,
+      "loss": 0.2208,
+      "step": 9090
+    },
+    {
+      "epoch": 0.9245148836736767,
+      "grad_norm": 3.546875,
+      "learning_rate": 7.140419154876372e-07,
+      "loss": 0.184,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9255308340952961,
+      "grad_norm": 2.640625,
+      "learning_rate": 6.950397136941872e-07,
+      "loss": 0.2097,
+      "step": 9110
+    },
+    {
+      "epoch": 0.9265467845169155,
+      "grad_norm": 3.671875,
+      "learning_rate": 6.762902140174888e-07,
+      "loss": 0.19,
+      "step": 9120
+    },
+    {
+      "epoch": 0.927562734938535,
+      "grad_norm": 2.03125,
+      "learning_rate": 6.577936113987437e-07,
+      "loss": 0.1427,
+      "step": 9130
+    },
+    {
+      "epoch": 0.9285786853601544,
+      "grad_norm": 2.1875,
+      "learning_rate": 6.395500981497577e-07,
+      "loss": 0.2116,
+      "step": 9140
+    },
+    {
+      "epoch": 0.9295946357817738,
+      "grad_norm": 5.25,
+      "learning_rate": 6.215598639509185e-07,
+      "loss": 0.2384,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9306105862033933,
+      "grad_norm": 3.953125,
+      "learning_rate": 6.038230958492403e-07,
+      "loss": 0.2406,
+      "step": 9160
+    },
+    {
+      "epoch": 0.9316265366250127,
+      "grad_norm": 4.09375,
+      "learning_rate": 5.863399782564199e-07,
+      "loss": 0.1889,
+      "step": 9170
+    },
+    {
+      "epoch": 0.9326424870466321,
+      "grad_norm": 3.109375,
+      "learning_rate": 5.691106929469004e-07,
+      "loss": 0.1916,
+      "step": 9180
+    },
+    {
+      "epoch": 0.9336584374682515,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.521354190560102e-07,
+      "loss": 0.1752,
+      "step": 9190
+    },
+    {
+      "epoch": 0.934674387889871,
+      "grad_norm": 5.21875,
+      "learning_rate": 5.354143330780714e-07,
+      "loss": 0.1779,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9356903383114904,
+      "grad_norm": 2.359375,
+      "learning_rate": 5.18947608864595e-07,
+      "loss": 0.1461,
+      "step": 9210
+    },
+    {
+      "epoch": 0.9367062887331098,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.027354176224353e-07,
+      "loss": 0.2565,
+      "step": 9220
+    },
+    {
+      "epoch": 0.9377222391547293,
+      "grad_norm": 4.5,
+      "learning_rate": 4.867779279120493e-07,
+      "loss": 0.2301,
+      "step": 9230
+    },
+    {
+      "epoch": 0.9387381895763487,
+      "grad_norm": 2.984375,
+      "learning_rate": 4.710753056457157e-07,
+      "loss": 0.1916,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9397541399979681,
+      "grad_norm": 4.40625,
+      "learning_rate": 4.556277140858267e-07,
+      "loss": 0.1808,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9407700904195875,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.404353138431766e-07,
+      "loss": 0.1552,
+      "step": 9260
+    },
+    {
+      "epoch": 0.941786040841207,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.254982628753096e-07,
+      "loss": 0.1995,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9428019912628264,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.108167164848575e-07,
+      "loss": 0.1495,
+      "step": 9280
+    },
+    {
+      "epoch": 0.9438179416844458,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.963908273179384e-07,
+      "loss": 0.1787,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9448338921060653,
+      "grad_norm": 0.8671875,
+      "learning_rate": 3.8222074536257144e-07,
+      "loss": 0.1742,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9458498425276847,
+      "grad_norm": 0.734375,
+      "learning_rate": 3.683066179470979e-07,
+      "loss": 0.1386,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9468657929493041,
+      "grad_norm": 2.796875,
+      "learning_rate": 3.5464858973868476e-07,
+      "loss": 0.1806,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9478817433709235,
+      "grad_norm": 2.734375,
+      "learning_rate": 3.4124680274177646e-07,
+      "loss": 0.1873,
+      "step": 9330
+    },
+    {
+      "epoch": 0.948897693792543,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.2810139629665393e-07,
+      "loss": 0.1745,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9499136442141624,
+      "grad_norm": 3.546875,
+      "learning_rate": 3.152125070779749e-07,
+      "loss": 0.2116,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9509295946357817,
+      "grad_norm": 5.5625,
+      "learning_rate": 3.0258026909334713e-07,
+      "loss": 0.2088,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9519455450574011,
+      "grad_norm": 4.5,
+      "learning_rate": 2.9020481368193795e-07,
+      "loss": 0.111,
+      "step": 9370
+    },
+    {
+      "epoch": 0.9529614954790206,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.7808626951310867e-07,
+      "loss": 0.2391,
+      "step": 9380
+    },
+    {
+      "epoch": 0.95397744590064,
+      "grad_norm": 2.6875,
+      "learning_rate": 2.662247625850822e-07,
+      "loss": 0.3217,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9549933963222594,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.5462041622362767e-07,
+      "loss": 0.1667,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9560093467438789,
+      "grad_norm": 0.8984375,
+      "learning_rate": 2.4327335108077773e-07,
+      "loss": 0.1709,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9570252971654983,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.3218368513357737e-07,
+      "loss": 0.1912,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9580412475871177,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.213515336828592e-07,
+      "loss": 0.1544,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9590571980087371,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.1077700935202836e-07,
+      "loss": 0.1806,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9600731484303566,
+      "grad_norm": 3.5625,
+      "learning_rate": 2.004602220859214e-07,
+      "loss": 0.152,
+      "step": 9450
+    },
+    {
+      "epoch": 0.961089098851976,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.9040127914963514e-07,
+      "loss": 0.1799,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9621050492735954,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.8060028512742188e-07,
+      "loss": 0.1811,
+      "step": 9470
+    },
+    {
+      "epoch": 0.9631209996952149,
+      "grad_norm": 5.875,
+      "learning_rate": 1.7105734192160717e-07,
+      "loss": 0.2012,
+      "step": 9480
+    },
+    {
+      "epoch": 0.9641369501168343,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.6177254875152647e-07,
+      "loss": 0.2129,
+      "step": 9490
+    },
+    {
+      "epoch": 0.9651529005384537,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.5274600215248736e-07,
+      "loss": 0.1498,
+      "step": 9500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9843,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 7,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoints/checkpoint-9500/training_args.bin b/checkpoints/checkpoint-9500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/checkpoints/checkpoint-9500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984
diff --git a/training_args.bin b/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6bb0a4ec10ab29ca1942aecdfa4212d352e373f1
--- /dev/null
+++ b/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
+size 4984