Upload trainer_state.json

trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 5.347593582887701,
-  "global_step": 1000,
+  "epoch": 16.0427807486631,
+  "global_step": 3000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -68,11 +68,135 @@
       "eval_samples_per_second": 18.484,
       "eval_steps_per_second": 4.646,
       "step": 1000
+    },
+    {
+      "epoch": 6.42,
+      "eval_accuracy": 0.9037433155080213,
+      "eval_loss": 0.8681638240814209,
+      "eval_recall": 0.7755102040816326,
+      "eval_runtime": 10.157,
+      "eval_samples_per_second": 18.411,
+      "eval_steps_per_second": 4.627,
+      "step": 1200
+    },
+    {
+      "epoch": 7.49,
+      "eval_accuracy": 0.8823529411764706,
+      "eval_loss": 0.9168533682823181,
+      "eval_recall": 0.6938775510204082,
+      "eval_runtime": 10.271,
+      "eval_samples_per_second": 18.207,
+      "eval_steps_per_second": 4.576,
+      "step": 1400
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.6631016042780753e-05,
+      "loss": 0.0555,
+      "step": 1500
+    },
+    {
+      "epoch": 8.56,
+      "eval_accuracy": 0.8556149732620321,
+      "eval_loss": 0.9982467293739319,
+      "eval_recall": 0.7346938775510204,
+      "eval_runtime": 10.134,
+      "eval_samples_per_second": 18.453,
+      "eval_steps_per_second": 4.638,
+      "step": 1600
+    },
+    {
+      "epoch": 9.63,
+      "eval_accuracy": 0.8663101604278075,
+      "eval_loss": 1.04860258102417,
+      "eval_recall": 0.7142857142857143,
+      "eval_runtime": 10.225,
+      "eval_samples_per_second": 18.289,
+      "eval_steps_per_second": 4.597,
+      "step": 1800
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2174688057041004e-05,
+      "loss": 0.0224,
+      "step": 2000
+    },
+    {
+      "epoch": 10.7,
+      "eval_accuracy": 0.8770053475935828,
+      "eval_loss": 1.0292185544967651,
+      "eval_recall": 0.673469387755102,
+      "eval_runtime": 10.362,
+      "eval_samples_per_second": 18.047,
+      "eval_steps_per_second": 4.536,
+      "step": 2000
+    },
+    {
+      "epoch": 11.76,
+      "eval_accuracy": 0.8663101604278075,
+      "eval_loss": 1.1402537822723389,
+      "eval_recall": 0.673469387755102,
+      "eval_runtime": 10.233,
+      "eval_samples_per_second": 18.274,
+      "eval_steps_per_second": 4.593,
+      "step": 2200
+    },
+    {
+      "epoch": 12.83,
+      "eval_accuracy": 0.8716577540106952,
+      "eval_loss": 1.1431100368499756,
+      "eval_recall": 0.673469387755102,
+      "eval_runtime": 10.226,
+      "eval_samples_per_second": 18.287,
+      "eval_steps_per_second": 4.596,
+      "step": 2400
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7718360071301247e-05,
+      "loss": 0.0001,
+      "step": 2500
+    },
+    {
+      "epoch": 13.9,
+      "eval_accuracy": 0.8770053475935828,
+      "eval_loss": 1.1465816497802734,
+      "eval_recall": 0.6530612244897959,
+      "eval_runtime": 10.254,
+      "eval_samples_per_second": 18.237,
+      "eval_steps_per_second": 4.584,
+      "step": 2600
+    },
+    {
+      "epoch": 14.97,
+      "eval_accuracy": 0.8716577540106952,
+      "eval_loss": 1.1757899522781372,
+      "eval_recall": 0.6530612244897959,
+      "eval_runtime": 10.346,
+      "eval_samples_per_second": 18.075,
+      "eval_steps_per_second": 4.543,
+      "step": 2800
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 2.32620320855615e-05,
+      "loss": 0.0,
+      "step": 3000
+    },
+    {
+      "epoch": 16.04,
+      "eval_accuracy": 0.8823529411764706,
+      "eval_loss": 1.1856350898742676,
+      "eval_recall": 0.6530612244897959,
+      "eval_runtime": 10.397,
+      "eval_samples_per_second": 17.986,
+      "eval_steps_per_second": 4.521,
+      "step": 3000
     }
   ],
   "max_steps": 5610,
   "num_train_epochs": 30,
-  "total_flos": 1049850814279680.0,
+  "total_flos": 3149026201829376.0,
   "trial_name": null,
   "trial_params": null
 }