🍻 cheers

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 4.0,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.01580428145825863,
-    "eval_runtime": 6.0348,
-    "eval_samples_per_second": 5.965,
-    "eval_steps_per_second": 0.829,
-    "train_loss": 0.09506800605191125,
-    "train_runtime": 143.0442,
-    "train_samples_per_second": 4.027,
-    "train_steps_per_second": 0.252
+    "eval_loss": 0.005373110994696617,
+    "eval_runtime": 6.0446,
+    "eval_samples_per_second": 5.956,
+    "eval_steps_per_second": 0.827,
+    "train_loss": 0.05584914175172647,
+    "train_runtime": 136.48,
+    "train_samples_per_second": 4.22,
+    "train_steps_per_second": 0.264
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.01580428145825863,
-    "eval_runtime": 6.0348,
-    "eval_samples_per_second": 5.965,
-    "eval_steps_per_second": 0.829
+    "eval_loss": 0.005373110994696617,
+    "eval_runtime": 6.0446,
+    "eval_samples_per_second": 5.956,
+    "eval_steps_per_second": 0.827
 }

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 4.0,
-    "train_loss": 0.09506800605191125,
-    "train_runtime": 143.0442,
-    "train_samples_per_second": 4.027,
-    "train_steps_per_second": 0.252
+    "train_loss": 0.05584914175172647,
+    "train_runtime": 136.48,
+    "train_samples_per_second": 4.22,
+    "train_steps_per_second": 0.264
 }

trainer_state.json

@@ -10,29 +10,29 @@
     {
       "epoch": 1.11,
       "learning_rate": 0.00014444444444444444,
-      "loss": 0.2916,
+      "loss": 0.182,
       "step": 10
     },
     {
       "epoch": 2.22,
       "learning_rate": 8.888888888888889e-05,
-      "loss": 0.0287,
+      "loss": 0.0101,
       "step": 20
     },
     {
       "epoch": 3.33,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.0147,
+      "loss": 0.0058,
       "step": 30
     },
     {
       "epoch": 4.0,
       "step": 36,
       "total_flos": 4.463538601800499e+16,
-      "train_loss": 0.09506800605191125,
-      "train_runtime": 143.0442,
-      "train_samples_per_second": 4.027,
-      "train_steps_per_second": 0.252
+      "train_loss": 0.05584914175172647,
+      "train_runtime": 136.48,
+      "train_samples_per_second": 4.22,
+      "train_steps_per_second": 0.264
     }
   ],
   "max_steps": 36,