Model save

all_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 5.19321114598441e+16,
-    "train_loss": 0.29687737567084177,
-    "train_runtime": 332.0278,
+    "train_loss": 0.3596490706716265,
+    "train_runtime": 336.8523,
     "train_samples": 9000,
-    "train_samples_per_second": 1.678,
-    "train_steps_per_second": 0.105
+    "train_samples_per_second": 1.657,
+    "train_steps_per_second": 0.104
 }

runs/Nov29_13-10-26_RLHF000/events.out.tfevents.1732885850.RLHF000.687743.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2581c00f4c1466bcf817413b6208387fb1ad5c48ea66e96a01ecf744c12266e5
-size 8570
+oid sha256:db5884ea7d51261793891b0ae9a0c23b9417bb6b1e0c193f6152171a47ee187d
+size 9184

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 5.19321114598441e+16,
-    "train_loss": 0.29687737567084177,
-    "train_runtime": 332.0278,
+    "train_loss": 0.3596490706716265,
+    "train_runtime": 336.8523,
     "train_samples": 9000,
-    "train_samples_per_second": 1.678,
-    "train_steps_per_second": 0.105
+    "train_samples_per_second": 1.657,
+    "train_steps_per_second": 0.104
 }

trainer_state.json

@@ -10,76 +10,76 @@
   "log_history": [
     {
       "epoch": 0.02857142857142857,
-      "grad_norm": 1.1770261526107788,
+      "grad_norm": 1.1929763555526733,
       "learning_rate": 5e-05,
-      "loss": 0.781,
+      "loss": 0.842,
       "step": 1
     },
     {
       "epoch": 0.14285714285714285,
-      "grad_norm": 0.7865971326828003,
+      "grad_norm": 0.5904757380485535,
       "learning_rate": 0.00019948693233918952,
-      "loss": 0.6954,
+      "loss": 0.7602,
       "step": 5
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 0.3400590717792511,
+      "grad_norm": 0.35242587327957153,
       "learning_rate": 0.00018207634412072764,
-      "loss": 0.4525,
+      "loss": 0.5166,
       "step": 10
     },
     {
       "epoch": 0.42857142857142855,
-      "grad_norm": 0.24245862662792206,
+      "grad_norm": 0.23204153776168823,
       "learning_rate": 0.00014403941515576344,
-      "loss": 0.306,
+      "loss": 0.3699,
       "step": 15
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 0.23055194318294525,
+      "grad_norm": 0.21575891971588135,
       "learning_rate": 9.493508311612874e-05,
-      "loss": 0.2153,
+      "loss": 0.2784,
       "step": 20
     },
     {
       "epoch": 0.7142857142857143,
-      "grad_norm": 0.25304117798805237,
+      "grad_norm": 0.27512410283088684,
       "learning_rate": 4.710359896730379e-05,
-      "loss": 0.1602,
+      "loss": 0.222,
       "step": 25
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 0.2968534231185913,
+      "grad_norm": 0.3192328214645386,
       "learning_rate": 1.2565338385541792e-05,
-      "loss": 0.1206,
+      "loss": 0.1813,
       "step": 30
     },
     {
       "epoch": 1.0,
-      "grad_norm": 0.11835439503192902,
+      "grad_norm": 0.11489653587341309,
       "learning_rate": 0.0,
-      "loss": 0.111,
+      "loss": 0.1727,
       "step": 35
     },
     {
       "epoch": 1.0,
-      "eval_loss": 0.11065573990345001,
-      "eval_runtime": 10.7768,
-      "eval_samples_per_second": 5.66,
-      "eval_steps_per_second": 0.371,
+      "eval_loss": 0.1722353994846344,
+      "eval_runtime": 11.0271,
+      "eval_samples_per_second": 5.623,
+      "eval_steps_per_second": 0.363,
       "step": 35
     },
     {
       "epoch": 1.0,
       "step": 35,
       "total_flos": 5.19321114598441e+16,
-      "train_loss": 0.29687737567084177,
-      "train_runtime": 332.0278,
-      "train_samples_per_second": 1.678,
-      "train_steps_per_second": 0.105
+      "train_loss": 0.3596490706716265,
+      "train_runtime": 336.8523,
+      "train_samples_per_second": 1.657,
+      "train_steps_per_second": 0.104
     }
   ],
   "logging_steps": 5,