End of training

trainer_state.json

@@ -1,69 +1,55 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.8,
+  "epoch": 4.0,
   "eval_steps": 500,
-  "global_step": 12,
+  "global_step": 4,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.32,
-      "grad_norm": 0.17948530614376068,
-      "learning_rate": 0.0001666666666666667,
-      "loss": 8.9023,
-      "step": 2
-    },
-    {
-      "epoch": 0.64,
-      "grad_norm": 0.33971741795539856,
-      "learning_rate": 0.00013333333333333334,
-      "loss": 8.7931,
-      "step": 4
-    },
-    {
-      "epoch": 0.96,
-      "grad_norm": 0.4960370361804962,
+      "epoch": 1.0,
+      "grad_norm": Infinity,
       "learning_rate": 0.0001,
-      "loss": 8.6895,
-      "step": 6
+      "loss": 3.4942,
+      "step": 1
     },
     {
-      "epoch": 1.16,
-      "grad_norm": 0.5553238391876221,
-      "learning_rate": 6.666666666666667e-05,
-      "loss": 5.3485,
-      "step": 8
+      "epoch": 2.0,
+      "grad_norm": 90.76869201660156,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.5171,
+      "step": 2
     },
     {
-      "epoch": 1.48,
-      "grad_norm": 0.7772231698036194,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 8.4708,
-      "step": 10
+      "epoch": 3.0,
+      "grad_norm": 52.46024703979492,
+      "learning_rate": 5e-05,
+      "loss": 3.0473,
+      "step": 3
     },
     {
-      "epoch": 1.8,
-      "grad_norm": 0.7240044474601746,
-      "learning_rate": 0.0,
-      "loss": 8.4396,
-      "step": 12
+      "epoch": 4.0,
+      "grad_norm": 12.389888763427734,
+      "learning_rate": 2.5e-05,
+      "loss": 2.8469,
+      "step": 4
     },
     {
-      "epoch": 1.8,
-      "step": 12,
-      "total_flos": 33316935156720.0,
-      "train_loss": 8.10729185740153,
-      "train_runtime": 84.0152,
-      "train_samples_per_second": 2.381,
-      "train_steps_per_second": 0.143
+      "epoch": 4.0,
+      "step": 4,
+      "total_flos": 333548660400.0,
+      "train_loss": 3.2263943552970886,
+      "train_runtime": 27.0103,
+      "train_samples_per_second": 0.74,
+      "train_steps_per_second": 0.148
     }
   ],
-  "logging_steps": 2,
-  "max_steps": 12,
+  "logging_steps": 1,
+  "max_steps": 4,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 2,
+  "num_train_epochs": 4,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -77,7 +63,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 33316935156720.0,
+  "total_flos": 333548660400.0,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null