End of training

all_results.json

@@ -3,13 +3,13 @@
     "eval_loss": NaN,
     "eval_mean_acc": 0.0,
     "eval_median_acc": 0.0,
-    "eval_runtime": 16.7609,
+    "eval_runtime": 16.2793,
     "eval_samples": 18477,
-    "eval_samples_per_second": 1102.386,
-    "eval_steps_per_second": 0.597,
+    "eval_samples_per_second": 1135.001,
+    "eval_steps_per_second": 0.614,
     "train_loss": 0.0,
-    "train_runtime": 109.5006,
+    "train_runtime": 124.2018,
     "train_samples": 100000,
-    "train_samples_per_second": 913.237,
-    "train_steps_per_second": 0.895
+    "train_samples_per_second": 805.142,
+    "train_steps_per_second": 0.789
 }

eval_results.json

@@ -3,8 +3,8 @@
     "eval_loss": NaN,
     "eval_mean_acc": 0.0,
     "eval_median_acc": 0.0,
-    "eval_runtime": 16.7609,
+    "eval_runtime": 16.2793,
     "eval_samples": 18477,
-    "eval_samples_per_second": 1102.386,
-    "eval_steps_per_second": 0.597
+    "eval_samples_per_second": 1135.001,
+    "eval_steps_per_second": 0.614
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
     "train_loss": 0.0,
-    "train_runtime": 109.5006,
+    "train_runtime": 124.2018,
     "train_samples": 100000,
-    "train_samples_per_second": 913.237,
-    "train_steps_per_second": 0.895
+    "train_samples_per_second": 805.142,
+    "train_steps_per_second": 0.789
 }

trainer_state.json

@@ -7,14 +7,24 @@
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
+    {
+      "epoch": 0.51,
+      "eval_loss": NaN,
+      "eval_mean_acc": 0.0,
+      "eval_median_acc": 0.0,
+      "eval_runtime": 16.276,
+      "eval_samples_per_second": 1135.228,
+      "eval_steps_per_second": 0.614,
+      "step": 50
+    },
     {
       "epoch": 1.0,
       "step": 98,
       "total_flos": 1.038017592e+16,
       "train_loss": 0.0,
-      "train_runtime": 109.5006,
-      "train_samples_per_second": 913.237,
-      "train_steps_per_second": 0.895
+      "train_runtime": 124.2018,
+      "train_samples_per_second": 805.142,
+      "train_steps_per_second": 0.789
     }
   ],
   "max_steps": 98,