Model save

all_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 0.0,
-    "train_loss": 0.8670984277358422,
-    "train_runtime": 947.0697,
+    "train_loss": 0.8588812213677627,
+    "train_runtime": 1105.0645,
     "train_samples": 2484,
-    "train_samples_per_second": 2.623,
-    "train_steps_per_second": 0.082
+    "train_samples_per_second": 2.248,
+    "train_steps_per_second": 0.071
 }

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bae4b6a575e3f2bd6dde76ec82c0cd73482a2b483f3e4fec95fc63ee323eee0c
+oid sha256:1d5c0146cd356b747c79d0c9ae9dee3ed8d79152bbc826a1c063c8f0204f69d8
 size 4877660776

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cc39669a1e3a4dca3d93e0dec032b5772d254de5e7a49a3e55eb1fd8f94d5213
+oid sha256:3684c155c5c59bd3ec33504b31fa193d29cdc86c5db0f3873e8446238d6ea06b
 size 4932751008

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5e26b214070fead02fc94b063910e2ab733096b53977d862b0c45a9d5b7068d4
+oid sha256:a9198cc18abf8235451670e34276c95cb3500675d4f98aea2d59c448ebb03678
 size 4330865200

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8dd3491c6cf2063cdc3b990c0d336a58807ed8871c33423554f80531ff55a48
+oid sha256:e80bbab28f6458d570c45119c9ab3c85f50d355a2315358d0cf61797d33dce08
 size 1089994880

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 0.0,
-    "train_loss": 0.8670984277358422,
-    "train_runtime": 947.0697,
+    "train_loss": 0.8588812213677627,
+    "train_runtime": 1105.0645,
     "train_samples": 2484,
-    "train_samples_per_second": 2.623,
-    "train_steps_per_second": 0.082
+    "train_samples_per_second": 2.248,
+    "train_steps_per_second": 0.071
 }

trainer_state.json

@@ -10,7 +10,7 @@
   "log_history": [
     {
       "epoch": 0.01282051282051282,
-      "grad_norm": 85.6017268244877,
+      "grad_norm": 85.24139006970277,
       "learning_rate": 6.25e-08,
       "logits/generated": -1.035823106765747,
       "logits/real": -0.7845579385757446,
@@ -25,117 +25,117 @@
     },
     {
       "epoch": 0.1282051282051282,
-      "grad_norm": 60.394755772470546,
+      "grad_norm": 57.78188477455482,
       "learning_rate": 4.857142857142857e-07,
-      "logits/generated": -1.0380977392196655,
-      "logits/real": -0.9431765675544739,
-      "logps/generated": -217.96856689453125,
-      "logps/real": -227.99403381347656,
-      "loss": 0.9082,
-      "rewards/accuracies": 0.5694444179534912,
-      "rewards/generated": 0.11325030773878098,
-      "rewards/margins": 0.022136474028229713,
-      "rewards/real": 0.13538677990436554,
+      "logits/generated": -1.00705885887146,
+      "logits/real": -0.9160953164100647,
+      "logps/generated": -217.96315002441406,
+      "logps/real": -228.15171813964844,
+      "loss": 0.9081,
+      "rewards/accuracies": 0.5,
+      "rewards/generated": 0.11379219591617584,
+      "rewards/margins": 0.00582469254732132,
+      "rewards/real": 0.11961688101291656,
       "step": 10
     },
     {
       "epoch": 0.2564102564102564,
-      "grad_norm": 51.89266128280823,
+      "grad_norm": 55.19183682235091,
       "learning_rate": 4.142857142857143e-07,
-      "logits/generated": -1.076645016670227,
-      "logits/real": -1.0400656461715698,
-      "logps/generated": -203.994140625,
-      "logps/real": -197.03224182128906,
-      "loss": 0.8977,
-      "rewards/accuracies": 0.5,
-      "rewards/generated": 0.5957759618759155,
-      "rewards/margins": 0.02595127746462822,
-      "rewards/real": 0.6217272877693176,
+      "logits/generated": -1.052150011062622,
+      "logits/real": -1.0221259593963623,
+      "logps/generated": -200.09579467773438,
+      "logps/real": -193.20050048828125,
+      "loss": 0.88,
+      "rewards/accuracies": 0.5625,
+      "rewards/generated": 0.9856117367744446,
+      "rewards/margins": 0.019290633499622345,
+      "rewards/real": 1.0049023628234863,
       "step": 20
     },
     {
       "epoch": 0.38461538461538464,
-      "grad_norm": 61.44237094531356,
+      "grad_norm": 70.19010436249343,
       "learning_rate": 3.4285714285714286e-07,
-      "logits/generated": -1.0620293617248535,
-      "logits/real": -0.9935398101806641,
-      "logps/generated": -198.7886199951172,
-      "logps/real": -196.684814453125,
-      "loss": 0.8674,
-      "rewards/accuracies": 0.6000000238418579,
-      "rewards/generated": 1.4610140323638916,
-      "rewards/margins": 0.0977163091301918,
-      "rewards/real": 1.5587302446365356,
+      "logits/generated": -1.1168638467788696,
+      "logits/real": -1.0286533832550049,
+      "logps/generated": -196.82545471191406,
+      "logps/real": -194.75421142578125,
+      "loss": 0.8597,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/generated": 1.657329797744751,
+      "rewards/margins": 0.09445972740650177,
+      "rewards/real": 1.7517893314361572,
       "step": 30
     },
     {
       "epoch": 0.5128205128205128,
-      "grad_norm": 54.65869490615913,
+      "grad_norm": 59.8434152735345,
       "learning_rate": 2.714285714285714e-07,
-      "logits/generated": -0.9623354077339172,
-      "logits/real": -0.8568245768547058,
-      "logps/generated": -176.11087036132812,
-      "logps/real": -208.48501586914062,
-      "loss": 0.8731,
-      "rewards/accuracies": 0.4749999940395355,
-      "rewards/generated": 2.0073351860046387,
-      "rewards/margins": 0.06225720793008804,
-      "rewards/real": 2.0695924758911133,
+      "logits/generated": -1.0656118392944336,
+      "logits/real": -0.9198936223983765,
+      "logps/generated": -174.37677001953125,
+      "logps/real": -206.28683471679688,
+      "loss": 0.8633,
+      "rewards/accuracies": 0.5249999761581421,
+      "rewards/generated": 2.180746555328369,
+      "rewards/margins": 0.10865961015224457,
+      "rewards/real": 2.2894062995910645,
       "step": 40
     },
     {
       "epoch": 0.6410256410256411,
-      "grad_norm": 47.595811149692466,
+      "grad_norm": 62.50654161583806,
       "learning_rate": 2e-07,
-      "logits/generated": -1.1244269609451294,
-      "logits/real": -0.9023059010505676,
-      "logps/generated": -168.40548706054688,
-      "logps/real": -198.1172332763672,
-      "loss": 0.8405,
-      "rewards/accuracies": 0.6625000238418579,
-      "rewards/generated": 2.258265733718872,
-      "rewards/margins": 0.20940211415290833,
-      "rewards/real": 2.467667818069458,
+      "logits/generated": -1.1591944694519043,
+      "logits/real": -0.9199365377426147,
+      "logps/generated": -167.31390380859375,
+      "logps/real": -196.71131896972656,
+      "loss": 0.8212,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/generated": 2.3674254417419434,
+      "rewards/margins": 0.24083653092384338,
+      "rewards/real": 2.608261823654175,
       "step": 50
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 58.1930213116206,
+      "grad_norm": 56.97503359063291,
       "learning_rate": 1.2857142857142855e-07,
-      "logits/generated": -1.0248687267303467,
-      "logits/real": -0.8792899250984192,
-      "logps/generated": -177.2668914794922,
-      "logps/real": -185.9062042236328,
-      "loss": 0.8628,
-      "rewards/accuracies": 0.6625000238418579,
-      "rewards/generated": 2.466890811920166,
-      "rewards/margins": 0.10949119180440903,
-      "rewards/real": 2.5763819217681885,
+      "logits/generated": -1.0447438955307007,
+      "logits/real": -0.9084192514419556,
+      "logps/generated": -174.6784210205078,
+      "logps/real": -183.24168395996094,
+      "loss": 0.8568,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/generated": 2.725738525390625,
+      "rewards/margins": 0.11709457635879517,
+      "rewards/real": 2.8428330421447754,
       "step": 60
     },
     {
       "epoch": 0.8974358974358975,
-      "grad_norm": 56.77078246951395,
+      "grad_norm": 70.79977219793189,
       "learning_rate": 5.714285714285714e-08,
-      "logits/generated": -0.9708479642868042,
-      "logits/real": -0.8643589019775391,
-      "logps/generated": -182.65744018554688,
-      "logps/real": -199.50527954101562,
-      "loss": 0.8634,
-      "rewards/accuracies": 0.6625000238418579,
-      "rewards/generated": 2.7308669090270996,
-      "rewards/margins": 0.10345491021871567,
-      "rewards/real": 2.834322214126587,
+      "logits/generated": -1.0091984272003174,
+      "logits/real": -0.8863734006881714,
+      "logps/generated": -181.19216918945312,
+      "logps/real": -198.0881805419922,
+      "loss": 0.85,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/generated": 2.8773930072784424,
+      "rewards/margins": 0.09863928705453873,
+      "rewards/real": 2.976032257080078,
       "step": 70
     },
     {
       "epoch": 1.0,
       "step": 78,
       "total_flos": 0.0,
-      "train_loss": 0.8670984277358422,
-      "train_runtime": 947.0697,
-      "train_samples_per_second": 2.623,
-      "train_steps_per_second": 0.082
+      "train_loss": 0.8588812213677627,
+      "train_runtime": 1105.0645,
+      "train_samples_per_second": 2.248,
+      "train_steps_per_second": 0.071
     }
   ],
   "logging_steps": 10,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b5e97a26f901c9804b8e5e5f571c4eed7eb3691f8b2c50acecaa23fec183fafd
+oid sha256:36d1a01787e4675e1cf15999aa9ea89097015ad5478ec8071e954863c2d2ee84
 size 6392