Training in progress, step 5100

adapter_config.json

@@ -14,13 +14,13 @@
   "r": 32,
   "revision": null,
   "target_modules": [
+    "v_proj",
+    "up_proj",
     "k_proj",
-    "o_proj",
+    "q_proj",
     "down_proj",
     "gate_proj",
-    "v_proj",
-    "q_proj",
-    "up_proj"
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:658b9b2dec5afef11956f93a69d1e5899dfaf7ec45314dbb9a4f4fe9a8d341ef
+oid sha256:0782e3c048e8ec06b53914deeacadabf30a533ab881541cd5057d49805c57014
 size 500897101

{checkpoint-4700 → checkpoint-5000}/adapter_model/adapter_config.json

@@ -14,13 +14,13 @@
   "r": 32,
   "revision": null,
   "target_modules": [
+    "k_proj",
+    "o_proj",
+    "down_proj",
     "gate_proj",
     "v_proj",
-    "k_proj",
     "q_proj",
-    "up_proj",
-    "o_proj",
-    "down_proj"
+    "up_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

{checkpoint-4700 → checkpoint-5000/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4935b48a5b5648a6247e6431de513dc0ecbb8eed09d9aa9f97291297c9b18387
+oid sha256:658b9b2dec5afef11956f93a69d1e5899dfaf7ec45314dbb9a4f4fe9a8d341ef
 size 500897101

{checkpoint-4700 → checkpoint-5100}/adapter_config.json

@@ -14,13 +14,13 @@
   "r": 32,
   "revision": null,
   "target_modules": [
-    "gate_proj",
     "v_proj",
+    "up_proj",
     "k_proj",
     "q_proj",
-    "up_proj",
-    "o_proj",
-    "down_proj"
+    "down_proj",
+    "gate_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

{checkpoint-4700/adapter_model → checkpoint-5100}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4935b48a5b5648a6247e6431de513dc0ecbb8eed09d9aa9f97291297c9b18387
+oid sha256:0782e3c048e8ec06b53914deeacadabf30a533ab881541cd5057d49805c57014
 size 500897101

{checkpoint-4700 → checkpoint-5100}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dd92fbc8afe889305be433d322c7392c60677f4339cc79fbf57befe5abd10a46
+oid sha256:6e547facfed70ca3ced4bba623ceed57a68ada7036877bcb0fe8abdc4206a71d
 size 1001752701

{checkpoint-4700 → checkpoint-5100}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:816bb5328119af370643bbe564517216967ae8d912623c25528531a17016932a
+oid sha256:93586c7c2f4af77dae57306cefbbba6e501b56255a6e8ab1e51526e93247ec0c
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:15fa31e015656d024cb46d941006d9b323f7b2e0b7baf2bdb8011085e9e1a4ce
+oid sha256:914cd17a4d5c2393130506ee74b380b7ebf49b994b8e1f7f9a5a5011789dcb9d
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3d8840a238febf1045ad1e12b89c195d910e4973eedb67701cebb360249b4ea6
+oid sha256:f79575017a903227eead6d8ec8c987f6c9f113b02cc3fdca1a05e6dfeb87c263
 size 27789

{checkpoint-4700 → checkpoint-5100}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:193e685cf4c4b39de4412f6cd58e847ef3f6d1c406011c2564c5b1e0abe05800
+oid sha256:e862437c6ef81136a36f5dcccf18047c8bf5b5ce5c4cb6a5de4068de8ac98fd4
 size 27789

{checkpoint-4700 → checkpoint-5100}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:35e9c04088652cf38eecbd9c05f50aacacbdcb03398a949564c0b578c73868ae
+oid sha256:2bd910734bd5ed0209fccf18e21529e6637d89e5d932f3863bdeb79c5a9955a0
 size 27789

{checkpoint-4700 → checkpoint-5100}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:59c25864a97d222e1a0369f7d27af105afc69ccc8217aea243f1f66bdab78418
+oid sha256:bd706d0aeecfb525da3ab5daa901fe7638c2e7f5d4cee63cd71b6ca026275bdd
 size 27789

{checkpoint-4700 → checkpoint-5100}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:420254978826d26db15b8bee5888481b0b5a8570d801a05c09a67a4d066af01c
+oid sha256:a3cc4068be271200f493c6e5933125535d31c00d193a77baaf617e995cb80113
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:42b6f1d581e7677f39b01a8a721f22856b33f3c5d9373a55ae0a99579fb78bb6
+oid sha256:134ec8578ce099215e557b373586d1334dcba1f9bc3678e21e8a11c8293273b5
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f829d440694541c2cd069f4d86a5309eaa7b9bdf210a37c2e2830c86b9682a0f
+oid sha256:2799584e6f3b97fb9250177cae1099c95a1a72f8c924828310d6e9c9f712a0ad
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b3374ed76502dd80903f9a87246528ada548fb77411537b19a9066136b0d14f3
+oid sha256:d2d851757976cd5436fb4f7e4f6b90e86fd7eb7fd174ef2f06b786fab4c8b687
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:38cf2161ae208051726734a6c053fea2ee1a9f80f9e51c0b04c582898e641023
+oid sha256:0944f297f8c989eaaaa8748e1b4a866de5757e645fa13d826211ad5f9bf81798
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:772a1c042d85e9a1000a7434b42d36f2c19fa33ac75a669e78dcf761c2e55882
+oid sha256:f4847f06e89ac0c33e93f4108e29bba8a6a11b57185752e5a61d1d159db7176b
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eed50f21762056e7e237d8dee609e0e041246c7be788a84a103bfe0e1349ea57
+oid sha256:32e20dcbaf8912d2c4ea76cd78324e384e6e3f340b552198fd77b189a7f80400
 size 27772

{checkpoint-4700 → checkpoint-5100}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dc52a4e66f1e0d427c31e689c87eaa325a10f1d35cb760342846084bc1bd5fbf
+oid sha256:6299cbf9ce668897a91d6363b1369a182cddf5e47a03415dbcb196480ea98de3
 size 27772

{checkpoint-4700 → checkpoint-5100}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84ff3c12e295fb051fd849b3d694d8827518c71141847e02d3f60bb982a2a971
+oid sha256:04c0158b55de3dc26efbfe9fb78f379b4093417c7720f3a9de19f86082d0caf3
 size 627

{checkpoint-4700 → checkpoint-5100}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.8224117875145405,
-  "global_step": 4700,
+  "epoch": 1.9775106630476929,
+  "global_step": 5100,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1142,11 +1142,115 @@
       "learning_rate": 6.702555108433461e-05,
       "loss": 0.789,
       "step": 4700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.606768635795574e-05,
+      "loss": 0.7902,
+      "step": 4725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.511332727445191e-05,
+      "loss": 0.7924,
+      "step": 4750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.416257243134747e-05,
+      "loss": 0.7957,
+      "step": 4775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.321552005380256e-05,
+      "loss": 0.7916,
+      "step": 4800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.22722679844652e-05,
+      "loss": 0.7867,
+      "step": 4825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.133291367336284e-05,
+      "loss": 0.7944,
+      "step": 4850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.039755416783457e-05,
+      "loss": 0.7982,
+      "step": 4875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.946628610250484e-05,
+      "loss": 0.7918,
+      "step": 4900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.853920568929996e-05,
+      "loss": 0.7921,
+      "step": 4925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.761640870750799e-05,
+      "loss": 0.7878,
+      "step": 4950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.669799049388375e-05,
+      "loss": 0.7901,
+      "step": 4975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.578404593279911e-05,
+      "loss": 0.7858,
+      "step": 5000
+    },
+    {
+      "epoch": 1.94,
+      "eval_loss": 0.807844877243042,
+      "eval_runtime": 59.586,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.889,
+      "step": 5000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.487466944644033e-05,
+      "loss": 0.7902,
+      "step": 5025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.3969954985052996e-05,
+      "loss": 0.7979,
+      "step": 5050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.306999601723579e-05,
+      "loss": 0.7931,
+      "step": 5075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.21748855202839e-05,
+      "loss": 0.7868,
+      "step": 5100
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.0239423325245276e+19,
+  "total_flos": 2.1963895120276226e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-4700 → checkpoint-5100}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4458bf553277b063d908ed9668b95abc04892c52d8d793f6b007433394d06f17
+oid sha256:b5cc18faa1420e425c2fed06bfe2dd967461487c15531bd94429b7a3c0c02a49
 size 4027

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:354c4769acaf21317d8ab58ba2dd133763f3bee1aec7488fb334e45f3e80ebc3
+oid sha256:b5cc18faa1420e425c2fed06bfe2dd967461487c15531bd94429b7a3c0c02a49
 size 4027