Training in progress, step 1000, checkpoint

checkpoint-1000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b5584d22e170a3bc826f867fdece5c5b075692fe7efb47af3caa80b33c7c3412
+oid sha256:0688e482e9f88ba8f20e835e53a1657f34866a68c245056f10f68866513382a3
 size 577789320

checkpoint-1000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e0bf87f524ee11d9b41c95e63143c877c14ba983d24f0bfe9f53910a0431d145
+oid sha256:69fec42678b24e36e3673f3d90b5466c0cd5b03dfdf7885abb9076f018576ea8
 size 1155772233

checkpoint-1000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b6db3e4c935ff5c70a271135dc95dfa374b2c15a57e22107e71497f425a082d1
+oid sha256:6782ee4581b95437cf5eb6f04357f25869f5c488a3f0d2d0eee869ccecf25d57
 size 14244

checkpoint-1000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6080dc3c4f31046c753abd07ed5f1a8449e11695e9772af37e8337628bc01d7f
+oid sha256:45070bc91d144eb7f163f351456cac0db81ee5e61490751953caec092c3e6eb7
 size 1064

checkpoint-1000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.45661377906799316,
+  "best_metric": 0.4438903331756592,
   "best_model_checkpoint": "mikhail-panzo/fil_b32_le5_s4000/checkpoint-1000",
-  "epoch": 22.22222222222222,
+  "epoch": 88.88888888888889,
   "eval_steps": 500,
   "global_step": 1000,
   "is_hyper_param_search": false,
@@ -9,166 +9,166 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.1111111111111112,
-      "grad_norm": 4.398683071136475,
-      "learning_rate": 2.4500000000000004e-07,
-      "loss": 0.8108,
+      "epoch": 4.444444444444445,
+      "grad_norm": 4.288165092468262,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 0.7989,
       "step": 50
     },
     {
-      "epoch": 2.2222222222222223,
-      "grad_norm": 3.726436138153076,
-      "learning_rate": 4.95e-07,
-      "loss": 0.792,
+      "epoch": 8.88888888888889,
+      "grad_norm": 2.4493463039398193,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.7701,
       "step": 100
     },
     {
-      "epoch": 3.3333333333333335,
-      "grad_norm": 2.0379676818847656,
-      "learning_rate": 7.450000000000001e-07,
-      "loss": 0.7413,
+      "epoch": 13.333333333333334,
+      "grad_norm": 1.3798907995224,
+      "learning_rate": 7.5e-07,
+      "loss": 0.7374,
       "step": 150
     },
     {
-      "epoch": 4.444444444444445,
-      "grad_norm": 2.0460364818573,
-      "learning_rate": 9.950000000000002e-07,
-      "loss": 0.7342,
+      "epoch": 17.77777777777778,
+      "grad_norm": 2.6173877716064453,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7101,
       "step": 200
     },
     {
-      "epoch": 5.555555555555555,
-      "grad_norm": 2.803539276123047,
-      "learning_rate": 1.2450000000000002e-06,
-      "loss": 0.7147,
+      "epoch": 22.22222222222222,
+      "grad_norm": 1.3756701946258545,
+      "learning_rate": 1.25e-06,
+      "loss": 0.6887,
       "step": 250
     },
     {
-      "epoch": 6.666666666666667,
-      "grad_norm": 1.9818241596221924,
-      "learning_rate": 1.495e-06,
-      "loss": 0.7072,
+      "epoch": 26.666666666666668,
+      "grad_norm": 1.0456843376159668,
+      "learning_rate": 1.5e-06,
+      "loss": 0.672,
       "step": 300
     },
     {
-      "epoch": 7.777777777777778,
-      "grad_norm": 3.081399917602539,
-      "learning_rate": 1.745e-06,
-      "loss": 0.6843,
+      "epoch": 31.11111111111111,
+      "grad_norm": 0.9518398642539978,
+      "learning_rate": 1.75e-06,
+      "loss": 0.6583,
       "step": 350
     },
     {
-      "epoch": 8.88888888888889,
-      "grad_norm": 2.4465785026550293,
-      "learning_rate": 1.9950000000000004e-06,
-      "loss": 0.657,
+      "epoch": 35.55555555555556,
+      "grad_norm": 1.305978775024414,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6258,
       "step": 400
     },
     {
-      "epoch": 10.0,
-      "grad_norm": 2.929323434829712,
-      "learning_rate": 2.245e-06,
-      "loss": 0.6597,
+      "epoch": 40.0,
+      "grad_norm": 0.8857623338699341,
+      "learning_rate": 2.25e-06,
+      "loss": 0.5835,
       "step": 450
     },
     {
-      "epoch": 11.11111111111111,
-      "grad_norm": 1.419311285018921,
-      "learning_rate": 2.4950000000000003e-06,
-      "loss": 0.6246,
+      "epoch": 44.44444444444444,
+      "grad_norm": 1.052182912826538,
+      "learning_rate": 2.5e-06,
+      "loss": 0.5539,
       "step": 500
     },
     {
-      "epoch": 11.11111111111111,
-      "eval_loss": 0.5375720262527466,
-      "eval_runtime": 9.2568,
-      "eval_samples_per_second": 17.177,
-      "eval_steps_per_second": 2.161,
+      "epoch": 44.44444444444444,
+      "eval_loss": 0.48496538400650024,
+      "eval_runtime": 9.4574,
+      "eval_samples_per_second": 16.812,
+      "eval_steps_per_second": 2.115,
       "step": 500
     },
     {
-      "epoch": 12.222222222222221,
-      "grad_norm": 2.3438374996185303,
-      "learning_rate": 2.7450000000000004e-06,
-      "loss": 0.6018,
+      "epoch": 48.888888888888886,
+      "grad_norm": 1.249458909034729,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 0.5387,
       "step": 550
     },
     {
-      "epoch": 13.333333333333334,
-      "grad_norm": 1.7441222667694092,
-      "learning_rate": 2.995e-06,
-      "loss": 0.5547,
+      "epoch": 53.333333333333336,
+      "grad_norm": 0.801243245601654,
+      "learning_rate": 3e-06,
+      "loss": 0.5344,
       "step": 600
     },
     {
-      "epoch": 14.444444444444445,
-      "grad_norm": 1.5592021942138672,
-      "learning_rate": 3.2450000000000003e-06,
-      "loss": 0.5466,
+      "epoch": 57.77777777777778,
+      "grad_norm": 1.1922223567962646,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 0.5259,
       "step": 650
     },
     {
-      "epoch": 15.555555555555555,
-      "grad_norm": 1.305444359779358,
-      "learning_rate": 3.495e-06,
-      "loss": 0.535,
+      "epoch": 62.22222222222222,
+      "grad_norm": 0.9428519606590271,
+      "learning_rate": 3.5e-06,
+      "loss": 0.5179,
       "step": 700
     },
     {
-      "epoch": 16.666666666666668,
-      "grad_norm": 2.2067477703094482,
-      "learning_rate": 3.745e-06,
-      "loss": 0.5343,
+      "epoch": 66.66666666666667,
+      "grad_norm": 0.755260705947876,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.5082,
       "step": 750
     },
     {
-      "epoch": 17.77777777777778,
-      "grad_norm": 1.667480707168579,
-      "learning_rate": 3.995000000000001e-06,
-      "loss": 0.5296,
+      "epoch": 71.11111111111111,
+      "grad_norm": 1.8587590456008911,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5096,
       "step": 800
     },
     {
-      "epoch": 18.88888888888889,
-      "grad_norm": 1.5465797185897827,
-      "learning_rate": 4.245e-06,
-      "loss": 0.5198,
+      "epoch": 75.55555555555556,
+      "grad_norm": 0.6854829788208008,
+      "learning_rate": 4.25e-06,
+      "loss": 0.4952,
       "step": 850
     },
     {
-      "epoch": 20.0,
-      "grad_norm": 2.760589361190796,
-      "learning_rate": 4.495e-06,
-      "loss": 0.5193,
+      "epoch": 80.0,
+      "grad_norm": 0.9088854789733887,
+      "learning_rate": 4.5e-06,
+      "loss": 0.4941,
       "step": 900
     },
     {
-      "epoch": 21.11111111111111,
-      "grad_norm": 1.4570050239562988,
-      "learning_rate": 4.745e-06,
-      "loss": 0.5124,
+      "epoch": 84.44444444444444,
+      "grad_norm": 0.8932746648788452,
+      "learning_rate": 4.75e-06,
+      "loss": 0.4936,
       "step": 950
     },
     {
-      "epoch": 22.22222222222222,
-      "grad_norm": 2.1914265155792236,
-      "learning_rate": 4.9950000000000005e-06,
-      "loss": 0.5034,
+      "epoch": 88.88888888888889,
+      "grad_norm": 0.9161014556884766,
+      "learning_rate": 5e-06,
+      "loss": 0.4867,
       "step": 1000
     },
     {
-      "epoch": 22.22222222222222,
-      "eval_loss": 0.45661377906799316,
-      "eval_runtime": 9.2158,
-      "eval_samples_per_second": 17.253,
-      "eval_steps_per_second": 2.17,
+      "epoch": 88.88888888888889,
+      "eval_loss": 0.4438903331756592,
+      "eval_runtime": 9.2831,
+      "eval_samples_per_second": 17.128,
+      "eval_steps_per_second": 2.154,
       "step": 1000
     }
   ],
   "logging_steps": 50,
   "max_steps": 4000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 89,
+  "num_train_epochs": 364,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -182,7 +182,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 6889857259725792.0,
+  "total_flos": 2.752618142335248e+16,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

checkpoint-1000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:05b7ecdded947540b9670cedf7cfe87afc7199b237efb642b915ffbb505d51dc
+oid sha256:2de69a241d012253ce4b1526ecd0f9fd0f7dfbedfea74bfff08144d42d9f523a
 size 5304