Upload trainer_state.json with huggingface_hub

trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.576271186440678,
+  "epoch": 1.8022598870056497,
   "eval_steps": 10,
-  "global_step": 140,
+  "global_step": 160,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -302,6 +302,48 @@
       "eval_samples_per_second": 4.333,
       "eval_steps_per_second": 1.083,
       "step": 140
+    },
+    {
+      "entropy": 6.287807840108871,
+      "epoch": 1.689265536723164,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.372093023255814e-05,
+      "loss": 6.183982086181641,
+      "mean_token_accuracy": 0.2382544383406639,
+      "num_tokens": 2527248.0,
+      "step": 150
+    },
+    {
+      "epoch": 1.689265536723164,
+      "eval_entropy": 6.602698942025502,
+      "eval_loss": 6.374426364898682,
+      "eval_mean_token_accuracy": 0.20121282618492842,
+      "eval_num_tokens": 2527248.0,
+      "eval_runtime": 43.9903,
+      "eval_samples_per_second": 4.365,
+      "eval_steps_per_second": 1.091,
+      "step": 150
+    },
+    {
+      "entropy": 6.482830649614334,
+      "epoch": 1.8022598870056497,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.2093023255813955e-05,
+      "loss": 6.326276779174805,
+      "mean_token_accuracy": 0.2152696281671524,
+      "num_tokens": 2704860.0,
+      "step": 160
+    },
+    {
+      "epoch": 1.8022598870056497,
+      "eval_entropy": 6.604644636313121,
+      "eval_loss": 6.371755123138428,
+      "eval_mean_token_accuracy": 0.20118677647163472,
+      "eval_num_tokens": 2704860.0,
+      "eval_runtime": 44.2676,
+      "eval_samples_per_second": 4.337,
+      "eval_steps_per_second": 1.084,
+      "step": 160
     }
   ],
   "logging_steps": 10,
@@ -321,7 +363,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.6869770537105203e+17,
+  "total_flos": 1.931435674847908e+17,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null