End of training

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 1.0,
+    "eval_logits/chosen": -2.59375,
+    "eval_logits/rejected": -2.625,
+    "eval_logps/chosen": -338.0,
+    "eval_logps/rejected": -664.0,
+    "eval_loss": 0.2634149193763733,
+    "eval_rewards/accuracies": 0.8849018216133118,
+    "eval_rewards/chosen": -2.0625,
+    "eval_rewards/margins": 3.390625,
+    "eval_rewards/rejected": -5.4375,
+    "eval_runtime": 2667.9916,
+    "eval_samples": 95033,
+    "eval_samples_per_second": 35.417,
+    "eval_steps_per_second": 0.554,
     "total_flos": 0.0,
     "train_loss": 0.5075161390510498,
     "train_runtime": 153689.2585,

config.json

@@ -31,6 +31,6 @@
   "tie_word_embeddings": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.46.0",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 128256
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -2.59375,
+    "eval_logits/rejected": -2.625,
+    "eval_logps/chosen": -338.0,
+    "eval_logps/rejected": -664.0,
+    "eval_loss": 0.2634149193763733,
+    "eval_rewards/accuracies": 0.8849018216133118,
+    "eval_rewards/chosen": -2.0625,
+    "eval_rewards/margins": 3.390625,
+    "eval_rewards/rejected": -5.4375,
+    "eval_runtime": 2667.9916,
+    "eval_samples": 95033,
+    "eval_samples_per_second": 35.417,
+    "eval_steps_per_second": 0.554
+}