Model save

all_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 0.992,
     "total_flos": 0.0,
-    "train_loss": 0.7529569604704457,
-    "train_runtime": 771.983,
+    "train_loss": 0.7647512930054818,
+    "train_runtime": 925.9529,
     "train_samples": 1992,
-    "train_samples_per_second": 2.58,
-    "train_steps_per_second": 0.08
+    "train_samples_per_second": 2.151,
+    "train_steps_per_second": 0.067
 }

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41bfc87b2d8a847603ebea51bc7b25a7613debce011bcfbe088bfa496c174c9a
+oid sha256:708727674b50dc5363238c8722b7a99c5425bef018d70c004456157fbf402eaa
 size 4877660776

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0124ffe62940ff4d539444bb6b7a81f15de9221129dfbf3d3345bde87bc06682
+oid sha256:cc1ab5bcdf2b1b4cb86ced3c5e795d5a159b13b7ec3915cfea539ef8d546ffc3
 size 4932751008

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e995cd19230e32acb39d34266352b974860b4fb26908e29bb7a503b901bb78d4
+oid sha256:b54db3cdfc585ca1216e80487da4b7662f396502f0ff0d350a6b7ef0bc0d9eef
 size 4330865200

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:83e85406baa864bc5a96061ef5ab1e6119a0bac267d985a72e6e80184110e161
+oid sha256:e156cfe6001cbfc0267d383dfdfd04eac8730b8f861d1273664ef256342d96a1
 size 1089994880

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 0.992,
     "total_flos": 0.0,
-    "train_loss": 0.7529569604704457,
-    "train_runtime": 771.983,
+    "train_loss": 0.7647512930054818,
+    "train_runtime": 925.9529,
     "train_samples": 1992,
-    "train_samples_per_second": 2.58,
-    "train_steps_per_second": 0.08
+    "train_samples_per_second": 2.151,
+    "train_steps_per_second": 0.067
 }

trainer_state.json

@@ -10,7 +10,7 @@
   "log_history": [
     {
       "epoch": 0.016,
-      "grad_norm": 90.30108238635525,
+      "grad_norm": 90.36061761169043,
       "learning_rate": 7.142857142857142e-08,
       "logits/generated": -0.7607710957527161,
       "logits/real": -0.3239991366863251,
@@ -25,102 +25,102 @@
     },
     {
       "epoch": 0.16,
-      "grad_norm": 81.98515918889403,
+      "grad_norm": 80.57778361752598,
       "learning_rate": 4.727272727272727e-07,
-      "logits/generated": -0.9451051354408264,
-      "logits/real": -0.6303450465202332,
-      "logps/generated": -261.0451354980469,
-      "logps/real": -270.76763916015625,
-      "loss": 0.9227,
-      "rewards/accuracies": 0.625,
-      "rewards/generated": 0.056389763951301575,
-      "rewards/margins": 0.14121507108211517,
-      "rewards/real": 0.19760483503341675,
+      "logits/generated": -0.9439048767089844,
+      "logits/real": -0.6272310018539429,
+      "logps/generated": -260.8728942871094,
+      "logps/real": -270.6895446777344,
+      "loss": 0.9236,
+      "rewards/accuracies": 0.5972222089767456,
+      "rewards/generated": 0.07361925393342972,
+      "rewards/margins": 0.1317952275276184,
+      "rewards/real": 0.20541447401046753,
       "step": 10
     },
     {
       "epoch": 0.32,
-      "grad_norm": 174.7113626218452,
+      "grad_norm": 184.2526206474729,
       "learning_rate": 3.818181818181818e-07,
-      "logits/generated": -0.9395920634269714,
-      "logits/real": -0.5305559635162354,
-      "logps/generated": -255.8331756591797,
-      "logps/real": -262.07177734375,
-      "loss": 0.7746,
-      "rewards/accuracies": 0.8125,
-      "rewards/generated": 0.9409279823303223,
-      "rewards/margins": 0.9559415578842163,
-      "rewards/real": 1.896869421005249,
+      "logits/generated": -0.9683843851089478,
+      "logits/real": -0.5614625215530396,
+      "logps/generated": -255.9654998779297,
+      "logps/real": -262.6753845214844,
+      "loss": 0.7753,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/generated": 0.9276968240737915,
+      "rewards/margins": 0.9088083505630493,
+      "rewards/real": 1.8365051746368408,
       "step": 20
     },
     {
       "epoch": 0.48,
-      "grad_norm": 59.116404789366975,
+      "grad_norm": 58.389369897312676,
       "learning_rate": 2.909090909090909e-07,
-      "logits/generated": -0.9359496831893921,
-      "logits/real": -0.592766523361206,
-      "logps/generated": -248.56893920898438,
-      "logps/real": -255.5909881591797,
-      "loss": 0.7096,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/generated": 1.2769505977630615,
-      "rewards/margins": 0.9968850016593933,
-      "rewards/real": 2.2738356590270996,
+      "logits/generated": -1.0661559104919434,
+      "logits/real": -0.7151978015899658,
+      "logps/generated": -250.697509765625,
+      "logps/real": -258.60211181640625,
+      "loss": 0.7304,
+      "rewards/accuracies": 0.6875,
+      "rewards/generated": 1.0640974044799805,
+      "rewards/margins": 0.9086271524429321,
+      "rewards/real": 1.9727245569229126,
       "step": 30
     },
     {
       "epoch": 0.64,
-      "grad_norm": 92.60746313254718,
+      "grad_norm": 131.96437613532245,
       "learning_rate": 2e-07,
-      "logits/generated": -0.8351278305053711,
-      "logits/real": -0.5679959058761597,
-      "logps/generated": -251.27236938476562,
-      "logps/real": -241.02664184570312,
-      "loss": 0.6859,
-      "rewards/accuracies": 0.8500000238418579,
-      "rewards/generated": 1.4600062370300293,
-      "rewards/margins": 1.1287152767181396,
-      "rewards/real": 2.588721752166748,
+      "logits/generated": -0.9631511569023132,
+      "logits/real": -0.6713980436325073,
+      "logps/generated": -252.99539184570312,
+      "logps/real": -243.7881317138672,
+      "loss": 0.7048,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/generated": 1.2877063751220703,
+      "rewards/margins": 1.0248647928237915,
+      "rewards/real": 2.3125712871551514,
       "step": 40
     },
     {
       "epoch": 0.8,
-      "grad_norm": 85.79598787282833,
+      "grad_norm": 351.71434098117106,
       "learning_rate": 1.0909090909090908e-07,
-      "logits/generated": -0.9641457796096802,
-      "logits/real": -0.6498032808303833,
-      "logps/generated": -250.719482421875,
-      "logps/real": -241.3539581298828,
-      "loss": 0.6968,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/generated": 1.4068793058395386,
-      "rewards/margins": 1.1703944206237793,
-      "rewards/real": 2.5772736072540283,
+      "logits/generated": -1.0679857730865479,
+      "logits/real": -0.7241894602775574,
+      "logps/generated": -252.9150848388672,
+      "logps/real": -244.3249053955078,
+      "loss": 0.7103,
+      "rewards/accuracies": 0.7749999761581421,
+      "rewards/generated": 1.1873204708099365,
+      "rewards/margins": 1.0928575992584229,
+      "rewards/real": 2.2801780700683594,
       "step": 50
     },
     {
       "epoch": 0.96,
-      "grad_norm": 68.73888204600256,
+      "grad_norm": 102.7524068481551,
       "learning_rate": 1.818181818181818e-08,
-      "logits/generated": -0.8538883924484253,
-      "logits/real": -0.608859658241272,
-      "logps/generated": -255.9438934326172,
-      "logps/real": -255.7367706298828,
-      "loss": 0.6938,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/generated": 1.5117708444595337,
-      "rewards/margins": 1.1012026071548462,
-      "rewards/real": 2.61297345161438,
+      "logits/generated": -0.8544826507568359,
+      "logits/real": -0.6093840003013611,
+      "logps/generated": -257.0255432128906,
+      "logps/real": -257.3138122558594,
+      "loss": 0.7126,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/generated": 1.4036052227020264,
+      "rewards/margins": 1.051663875579834,
+      "rewards/real": 2.4552693367004395,
       "step": 60
     },
     {
       "epoch": 0.992,
       "step": 62,
       "total_flos": 0.0,
-      "train_loss": 0.7529569604704457,
-      "train_runtime": 771.983,
-      "train_samples_per_second": 2.58,
-      "train_steps_per_second": 0.08
+      "train_loss": 0.7647512930054818,
+      "train_runtime": 925.9529,
+      "train_samples_per_second": 2.151,
+      "train_steps_per_second": 0.067
     }
   ],
   "logging_steps": 10,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8cd49007af7a5717ec35e8140174a325191a6be53da07a59c00045fe784b3650
+oid sha256:5b763a587e39fee66e457f0de0320ba8e9a9dc5b927255da7b1324a7772b90da
 size 6392