Model save

README.md

@@ -32,7 +32,7 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 5e-07
+- learning_rate: 4e-07
 - train_batch_size: 4
 - eval_batch_size: 4
 - seed: 42
@@ -52,7 +52,7 @@ The following hyperparameters were used during training:
 
 ### Framework versions
 
-- Transformers 4.39.3
+- Transformers 4.41.1
 - Pytorch 2.1.2+cu118
 - Datasets 2.16.1
-- Tokenizers 0.15.2
+- Tokenizers 0.19.1

all_results.json

@@ -1,8 +1,9 @@
 {
-    "epoch": 1.0,
-    "train_loss": 0.4513903107398596,
-    "train_runtime": 5854.463,
+    "epoch": 0.9980806142034548,
+    "total_flos": 0.0,
+    "train_loss": 0.3119094243416419,
+    "train_runtime": 6443.086,
     "train_samples": 50000,
-    "train_samples_per_second": 8.54,
-    "train_steps_per_second": 0.067
+    "train_samples_per_second": 7.76,
+    "train_steps_per_second": 0.061
 }

config.json

@@ -20,7 +20,7 @@
   "sliding_window": 4096,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.39.3",
+  "transformers_version": "4.41.1",
   "use_cache": false,
   "vocab_size": 32000
 }

generation_config.json

@@ -2,5 +2,5 @@
   "_from_model_config": true,
   "bos_token_id": 1,
   "eos_token_id": 2,
-  "transformers_version": "4.39.3"
+  "transformers_version": "4.41.1"
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:04f2436f631684bd85d9ee0f015c2a6d18e79bd1c5a33dd8503659b12524b028
+oid sha256:5bc88b0288a40faf21a1e24a96384ccf683d38e8246d3b0cab22d31c505d8b9f
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:70de4abb0b18d600edfa17cb8203a64d82cb5cd93011b4c04bcdbe4e85b25295
+oid sha256:c430b926ebd01f65609ab319292fa842772b87fc92486924b50fdd7065be6155
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3f381addd443a7edabe9d8200e1356682fb04f61fecc2810526dc1a4c51d7868
+oid sha256:4d5049e6472b9a1956a056ed3a8a66408f55efa3d2ea2bd03073317eac4eec19
 size 4540516344

runs/Jun22_08-58-28_n136-082-130/events.out.tfevents.1719017941.n136-082-130.2587327.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:322059a7357c1ed3b22a6955d1cc16bf4648a1464ced2249bb3c51a70d6dfacf
-size 26299
+oid sha256:0ae7f710e69ad55192397299709b08f7cf7e04bccf24313d743eb75ad7f03d42
+size 32845

tokenizer.json

@@ -134,6 +134,7 @@
     "end_of_word_suffix": null,
     "fuse_unk": true,
     "byte_fallback": true,
+    "ignore_merges": false,
     "vocab": {
       "<unk>": 0,
       "<s>": 1,

train_results.json

@@ -1,8 +1,9 @@
 {
-    "epoch": 1.0,
-    "train_loss": 0.4513903107398596,
-    "train_runtime": 5854.463,
+    "epoch": 0.9980806142034548,
+    "total_flos": 0.0,
+    "train_loss": 0.3119094243416419,
+    "train_runtime": 6443.086,
     "train_samples": 50000,
-    "train_samples_per_second": 8.54,
-    "train_steps_per_second": 0.067
+    "train_samples_per_second": 7.76,
+    "train_steps_per_second": 0.061
 }

trainer_state.json

@@ -9,14 +9,14 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0,
-      "grad_norm": 1292.2103490533002,
-      "learning_rate": 1.282051282051282e-08,
+      "epoch": 0.0025591810620601407,
+      "grad_norm": 709.5459154500938,
+      "learning_rate": 1.0256410256410256e-08,
       "logits/chosen": -2.5583817958831787,
       "logits/rejected": -2.4487552642822266,
       "logps/chosen": -258.1644592285156,
       "logps/rejected": -216.25729370117188,
-      "loss": 0.6964,
+      "loss": 0.6931,
       "rewards/accuracies": 0.0,
       "rewards/chosen": 0.0,
       "rewards/margins": 0.0,
@@ -24,598 +24,598 @@
       "step": 1
     },
     {
-      "epoch": 0.03,
-      "grad_norm": 1261.1198733310573,
-      "learning_rate": 1.2820512820512818e-07,
-      "logits/chosen": -2.6061007976531982,
-      "logits/rejected": -2.553147315979004,
-      "logps/chosen": -267.5506591796875,
-      "logps/rejected": -217.63583374023438,
-      "loss": 0.7045,
-      "rewards/accuracies": 0.4375,
-      "rewards/chosen": 0.012439604848623276,
-      "rewards/margins": 0.01009546872228384,
-      "rewards/rejected": 0.002344133099541068,
+      "epoch": 0.025591810620601407,
+      "grad_norm": 684.196668594934,
+      "learning_rate": 1.0256410256410255e-07,
+      "logits/chosen": -2.605860471725464,
+      "logits/rejected": -2.5528433322906494,
+      "logps/chosen": -267.56048583984375,
+      "logps/rejected": -217.59442138671875,
+      "loss": 0.7027,
+      "rewards/accuracies": 0.4166666567325592,
+      "rewards/chosen": 0.0020057493820786476,
+      "rewards/margins": -0.020002255216240883,
+      "rewards/rejected": 0.022008005529642105,
       "step": 10
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 853.2707411908161,
-      "learning_rate": 2.5641025641025636e-07,
-      "logits/chosen": -2.629751443862915,
-      "logits/rejected": -2.5669989585876465,
-      "logps/chosen": -260.5412292480469,
-      "logps/rejected": -207.0039825439453,
-      "loss": 0.5192,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": 0.6866833567619324,
-      "rewards/margins": 0.6415502429008484,
-      "rewards/rejected": 0.045133065432310104,
+      "epoch": 0.05118362124120281,
+      "grad_norm": 515.3907221989738,
+      "learning_rate": 2.051282051282051e-07,
+      "logits/chosen": -2.6285552978515625,
+      "logits/rejected": -2.566141128540039,
+      "logps/chosen": -260.791259765625,
+      "logps/rejected": -207.073486328125,
+      "loss": 0.5875,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.2564736604690552,
+      "rewards/margins": 0.2661496698856354,
+      "rewards/rejected": -0.00967598520219326,
       "step": 20
     },
     {
-      "epoch": 0.08,
-      "grad_norm": 1142.6320871817904,
-      "learning_rate": 3.8461538461538463e-07,
-      "logits/chosen": -2.6441783905029297,
-      "logits/rejected": -2.5700392723083496,
-      "logps/chosen": -251.12313842773438,
-      "logps/rejected": -198.34071350097656,
-      "loss": 0.3383,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 3.621673583984375,
-      "rewards/margins": 3.1466643810272217,
-      "rewards/rejected": 0.47500935196876526,
+      "epoch": 0.07677543186180422,
+      "grad_norm": 608.3322300561331,
+      "learning_rate": 3.076923076923077e-07,
+      "logits/chosen": -2.6465039253234863,
+      "logits/rejected": -2.5724854469299316,
+      "logps/chosen": -251.90835571289062,
+      "logps/rejected": -198.9151611328125,
+      "loss": 0.3737,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/chosen": 1.6194490194320679,
+      "rewards/margins": 1.6427723169326782,
+      "rewards/rejected": -0.02332335151731968,
       "step": 30
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 498.4795422948633,
-      "learning_rate": 4.99989986344963e-07,
-      "logits/chosen": -2.6392924785614014,
-      "logits/rejected": -2.561156749725342,
-      "logps/chosen": -243.8309783935547,
-      "logps/rejected": -192.75845336914062,
-      "loss": 0.3207,
-      "rewards/accuracies": 0.8500000238418579,
-      "rewards/chosen": 5.866530418395996,
-      "rewards/margins": 5.031473636627197,
-      "rewards/rejected": 0.8350569009780884,
+      "epoch": 0.10236724248240563,
+      "grad_norm": 286.23056215212245,
+      "learning_rate": 3.999919890759704e-07,
+      "logits/chosen": -2.650782823562622,
+      "logits/rejected": -2.571556329727173,
+      "logps/chosen": -244.2300567626953,
+      "logps/rejected": -193.5239715576172,
+      "loss": 0.3167,
+      "rewards/accuracies": 0.793749988079071,
+      "rewards/chosen": 3.0596303939819336,
+      "rewards/margins": 2.9784700870513916,
+      "rewards/rejected": 0.08116073161363602,
       "step": 40
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 687.258084457276,
-      "learning_rate": 4.987893180827479e-07,
-      "logits/chosen": -2.6534440517425537,
-      "logits/rejected": -2.5842742919921875,
-      "logps/chosen": -258.9037170410156,
-      "logps/rejected": -203.41665649414062,
-      "loss": 0.3666,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 8.400110244750977,
-      "rewards/margins": 6.8410444259643555,
-      "rewards/rejected": 1.559065818786621,
+      "epoch": 0.12795905310300704,
+      "grad_norm": 355.8436814450708,
+      "learning_rate": 3.9903145446619833e-07,
+      "logits/chosen": -2.6738078594207764,
+      "logits/rejected": -2.601217031478882,
+      "logps/chosen": -257.42144775390625,
+      "logps/rejected": -202.82069396972656,
+      "loss": 0.3214,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 5.407862663269043,
+      "rewards/margins": 4.2437334060668945,
+      "rewards/rejected": 1.164129376411438,
       "step": 50
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 759.1516730607283,
-      "learning_rate": 4.955969343539162e-07,
-      "logits/chosen": -2.616115093231201,
-      "logits/rejected": -2.5453152656555176,
-      "logps/chosen": -262.9014892578125,
-      "logps/rejected": -209.2264404296875,
-      "loss": 0.3916,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 5.333600044250488,
-      "rewards/margins": 6.3847455978393555,
-      "rewards/rejected": -1.0511456727981567,
+      "epoch": 0.15355086372360843,
+      "grad_norm": 419.99201519842893,
+      "learning_rate": 3.9647754748313294e-07,
+      "logits/chosen": -2.6383204460144043,
+      "logits/rejected": -2.5659520626068115,
+      "logps/chosen": -259.7900695800781,
+      "logps/rejected": -207.2904510498047,
+      "loss": 0.3176,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/chosen": 4.518810272216797,
+      "rewards/margins": 4.13478422164917,
+      "rewards/rejected": 0.3840256929397583,
       "step": 60
     },
     {
-      "epoch": 0.18,
-      "grad_norm": 560.0502144214485,
-      "learning_rate": 4.90438392204474e-07,
-      "logits/chosen": -2.590355634689331,
-      "logits/rejected": -2.5212607383728027,
-      "logps/chosen": -292.03955078125,
-      "logps/rejected": -227.56900024414062,
-      "loss": 0.366,
-      "rewards/accuracies": 0.918749988079071,
-      "rewards/chosen": 5.28397798538208,
-      "rewards/margins": 7.424477577209473,
-      "rewards/rejected": -2.1404995918273926,
+      "epoch": 0.17914267434420986,
+      "grad_norm": 305.786716914146,
+      "learning_rate": 3.9235071376357917e-07,
+      "logits/chosen": -2.6047496795654297,
+      "logits/rejected": -2.5317835807800293,
+      "logps/chosen": -290.00616455078125,
+      "logps/rejected": -227.17697143554688,
+      "loss": 0.301,
+      "rewards/accuracies": 0.90625,
+      "rewards/chosen": 3.9522290229797363,
+      "rewards/margins": 4.945390701293945,
+      "rewards/rejected": -0.9931615591049194,
       "step": 70
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 1037.6404299504059,
-      "learning_rate": 4.83354989019146e-07,
-      "logits/chosen": -2.5414998531341553,
-      "logits/rejected": -2.468327283859253,
-      "logps/chosen": -260.0003356933594,
-      "logps/rejected": -203.8599090576172,
-      "loss": 0.3627,
+      "epoch": 0.20473448496481125,
+      "grad_norm": 512.2762241729247,
+      "learning_rate": 3.8668399121531677e-07,
+      "logits/chosen": -2.552280902862549,
+      "logits/rejected": -2.476268768310547,
+      "logps/chosen": -260.35906982421875,
+      "logps/rejected": -206.0688018798828,
+      "loss": 0.2615,
       "rewards/accuracies": 0.875,
-      "rewards/chosen": 6.802065372467041,
-      "rewards/margins": 7.507330894470215,
-      "rewards/rejected": -0.705264687538147,
+      "rewards/chosen": 3.5995476245880127,
+      "rewards/margins": 5.095798969268799,
+      "rewards/rejected": -1.4962517023086548,
       "step": 80
     },
     {
-      "epoch": 0.23,
-      "grad_norm": 1095.0164975833522,
-      "learning_rate": 4.7440343190975353e-07,
-      "logits/chosen": -2.5701346397399902,
-      "logits/rejected": -2.513249397277832,
-      "logps/chosen": -257.8319396972656,
-      "logps/rejected": -217.47695922851562,
-      "loss": 0.3529,
-      "rewards/accuracies": 0.84375,
-      "rewards/chosen": 3.5220108032226562,
-      "rewards/margins": 6.205338478088379,
-      "rewards/rejected": -2.683328151702881,
+      "epoch": 0.23032629558541268,
+      "grad_norm": 515.1909309248075,
+      "learning_rate": 3.795227455278028e-07,
+      "logits/chosen": -2.5692882537841797,
+      "logits/rejected": -2.510436534881592,
+      "logps/chosen": -256.6405334472656,
+      "logps/rejected": -217.69192504882812,
+      "loss": 0.264,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/chosen": 2.5523829460144043,
+      "rewards/margins": 4.150609016418457,
+      "rewards/rejected": -1.5982261896133423,
       "step": 90
     },
     {
-      "epoch": 0.26,
-      "grad_norm": 493.8561719761038,
-      "learning_rate": 4.6365538373900506e-07,
-      "logits/chosen": -2.624537944793701,
-      "logits/rejected": -2.5495262145996094,
-      "logps/chosen": -237.0422821044922,
-      "logps/rejected": -201.1056671142578,
-      "loss": 0.6296,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 4.613960266113281,
-      "rewards/margins": 6.805362701416016,
-      "rewards/rejected": -2.191401958465576,
+      "epoch": 0.2559181062060141,
+      "grad_norm": 587.7494413414474,
+      "learning_rate": 3.7092430699120403e-07,
+      "logits/chosen": -2.608997344970703,
+      "logits/rejected": -2.530529260635376,
+      "logps/chosen": -235.7121124267578,
+      "logps/rejected": -201.58700561523438,
+      "loss": 0.3994,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 3.2284069061279297,
+      "rewards/margins": 4.6865153312683105,
+      "rewards/rejected": -1.4581090211868286,
       "step": 100
     },
     {
-      "epoch": 0.28,
-      "grad_norm": 783.3339585792174,
-      "learning_rate": 4.5119688941406386e-07,
-      "logits/chosen": -2.6185028553009033,
-      "logits/rejected": -2.5373358726501465,
-      "logps/chosen": -258.2229919433594,
-      "logps/rejected": -209.95010375976562,
-      "loss": 0.5085,
+      "epoch": 0.28150991682661547,
+      "grad_norm": 383.9666636386822,
+      "learning_rate": 3.6095751153125107e-07,
+      "logits/chosen": -2.605412483215332,
+      "logits/rejected": -2.5209498405456543,
+      "logps/chosen": -257.01318359375,
+      "logps/rejected": -210.5634002685547,
+      "loss": 0.3274,
       "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 6.706292629241943,
-      "rewards/margins": 8.07982349395752,
-      "rewards/rejected": -1.3735301494598389,
+      "rewards/chosen": 4.330618381500244,
+      "rewards/margins": 5.40033483505249,
+      "rewards/rejected": -1.069716453552246,
       "step": 110
     },
     {
-      "epoch": 0.31,
-      "grad_norm": 1013.1425815966796,
-      "learning_rate": 4.3712768704277524e-07,
-      "logits/chosen": -2.5965781211853027,
-      "logits/rejected": -2.5263915061950684,
-      "logps/chosen": -263.1861572265625,
-      "logps/rejected": -208.8050079345703,
-      "loss": 0.4699,
-      "rewards/accuracies": 0.8812500238418579,
-      "rewards/chosen": 5.6486382484436035,
-      "rewards/margins": 7.843411445617676,
-      "rewards/rejected": -2.194772720336914,
+      "epoch": 0.30710172744721687,
+      "grad_norm": 452.32780946070125,
+      "learning_rate": 3.497021496342202e-07,
+      "logits/chosen": -2.5948054790496826,
+      "logits/rejected": -2.5218663215637207,
+      "logps/chosen": -261.63739013671875,
+      "logps/rejected": -209.2183074951172,
+      "loss": 0.3015,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": 3.912487030029297,
+      "rewards/margins": 5.338441371917725,
+      "rewards/rejected": -1.4259542226791382,
       "step": 120
     },
     {
-      "epoch": 0.33,
-      "grad_norm": 659.1716861853262,
-      "learning_rate": 4.2156040946718343e-07,
-      "logits/chosen": -2.571542263031006,
-      "logits/rejected": -2.502825975418091,
-      "logps/chosen": -252.36941528320312,
-      "logps/rejected": -197.37109375,
-      "loss": 0.4289,
+      "epoch": 0.3326935380678183,
+      "grad_norm": 325.92030560656247,
+      "learning_rate": 3.372483275737467e-07,
+      "logits/chosen": -2.574514389038086,
+      "logits/rejected": -2.505225896835327,
+      "logps/chosen": -250.41543579101562,
+      "logps/rejected": -196.6962127685547,
+      "loss": 0.2826,
       "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 4.440427303314209,
-      "rewards/margins": 7.714503288269043,
-      "rewards/rejected": -3.274075746536255,
+      "rewards/chosen": 3.443896532058716,
+      "rewards/margins": 4.925398826599121,
+      "rewards/rejected": -1.4815022945404053,
       "step": 130
     },
     {
-      "epoch": 0.36,
-      "grad_norm": 706.862930333198,
-      "learning_rate": 4.046196825665637e-07,
-      "logits/chosen": -2.5883898735046387,
-      "logits/rejected": -2.5185062885284424,
-      "logps/chosen": -271.88494873046875,
-      "logps/rejected": -218.22323608398438,
-      "loss": 0.4943,
-      "rewards/accuracies": 0.84375,
-      "rewards/chosen": 2.9622278213500977,
-      "rewards/margins": 7.161763668060303,
-      "rewards/rejected": -4.199534893035889,
+      "epoch": 0.3582853486884197,
+      "grad_norm": 286.3934639339848,
+      "learning_rate": 3.2369574605325094e-07,
+      "logits/chosen": -2.598271369934082,
+      "logits/rejected": -2.5281975269317627,
+      "logps/chosen": -268.22943115234375,
+      "logps/rejected": -216.2914581298828,
+      "loss": 0.3028,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": 3.4734299182891846,
+      "rewards/margins": 4.84061336517334,
+      "rewards/rejected": -1.3671828508377075,
       "step": 140
     },
     {
-      "epoch": 0.38,
-      "grad_norm": 715.990249793818,
-      "learning_rate": 3.864411275486261e-07,
-      "logits/chosen": -2.5699923038482666,
-      "logits/rejected": -2.502592086791992,
-      "logps/chosen": -264.418212890625,
-      "logps/rejected": -213.1693115234375,
-      "loss": 0.5111,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 5.556182384490967,
-      "rewards/margins": 8.052742004394531,
-      "rewards/rejected": -2.4965591430664062,
+      "epoch": 0.3838771593090211,
+      "grad_norm": 326.69658537939597,
+      "learning_rate": 3.091529020389009e-07,
+      "logits/chosen": -2.5794036388397217,
+      "logits/rejected": -2.5112969875335693,
+      "logps/chosen": -262.27850341796875,
+      "logps/rejected": -213.2576141357422,
+      "loss": 0.3047,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": 4.1566081047058105,
+      "rewards/margins": 5.587738990783691,
+      "rewards/rejected": -1.4311310052871704,
       "step": 150
     },
     {
-      "epoch": 0.41,
-      "grad_norm": 710.6520782907305,
-      "learning_rate": 3.671702752161759e-07,
-      "logits/chosen": -2.559586763381958,
-      "logits/rejected": -2.4897525310516357,
-      "logps/chosen": -245.1999969482422,
-      "logps/rejected": -198.19442749023438,
-      "loss": 0.4696,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 3.364131212234497,
-      "rewards/margins": 8.019769668579102,
-      "rewards/rejected": -4.655638694763184,
+      "epoch": 0.4094689699296225,
+      "grad_norm": 334.0349111221283,
+      "learning_rate": 2.937362201729407e-07,
+      "logits/chosen": -2.569304943084717,
+      "logits/rejected": -2.4969332218170166,
+      "logps/chosen": -241.8642120361328,
+      "logps/rejected": -196.44754028320312,
+      "loss": 0.2947,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 3.53686785697937,
+      "rewards/margins": 5.24989128112793,
+      "rewards/rejected": -1.7130241394042969,
       "step": 160
     },
     {
-      "epoch": 0.44,
-      "grad_norm": 865.8215089865342,
-      "learning_rate": 3.4696140090121375e-07,
-      "logits/chosen": -2.57694935798645,
-      "logits/rejected": -2.5136027336120605,
-      "logps/chosen": -266.91412353515625,
-      "logps/rejected": -211.6914825439453,
-      "loss": 0.397,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 3.399867296218872,
-      "rewards/margins": 8.182429313659668,
-      "rewards/rejected": -4.782561302185059,
+      "epoch": 0.4350607805502239,
+      "grad_norm": 385.1324260357516,
+      "learning_rate": 2.77569120720971e-07,
+      "logits/chosen": -2.5677335262298584,
+      "logits/rejected": -2.499415874481201,
+      "logps/chosen": -263.1123046875,
+      "logps/rejected": -209.7564239501953,
+      "loss": 0.2546,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 3.789729595184326,
+      "rewards/margins": 5.479173183441162,
+      "rewards/rejected": -1.6894439458847046,
       "step": 170
     },
     {
-      "epoch": 0.46,
-      "grad_norm": 658.5054936178984,
-      "learning_rate": 3.259762893935617e-07,
-      "logits/chosen": -2.6607277393341064,
-      "logits/rejected": -2.581068515777588,
-      "logps/chosen": -239.2282257080078,
-      "logps/rejected": -188.63772583007812,
-      "loss": 0.4691,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 2.5246458053588867,
-      "rewards/margins": 7.373200416564941,
-      "rewards/rejected": -4.848554611206055,
+      "epoch": 0.46065259117082535,
+      "grad_norm": 298.5381373642981,
+      "learning_rate": 2.6078103151484935e-07,
+      "logits/chosen": -2.632197856903076,
+      "logits/rejected": -2.5449578762054443,
+      "logps/chosen": -234.98526000976562,
+      "logps/rejected": -185.88897705078125,
+      "loss": 0.3076,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.5240702629089355,
+      "rewards/margins": 4.843334674835205,
+      "rewards/rejected": -1.3192641735076904,
       "step": 180
     },
     {
-      "epoch": 0.49,
-      "grad_norm": 746.921617266345,
-      "learning_rate": 3.0438293975154184e-07,
-      "logits/chosen": -2.6273646354675293,
-      "logits/rejected": -2.5594921112060547,
-      "logps/chosen": -263.73443603515625,
-      "logps/rejected": -207.8045654296875,
-      "loss": 0.4399,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 0.9775351285934448,
-      "rewards/margins": 8.604721069335938,
-      "rewards/rejected": -7.6271867752075195,
+      "epoch": 0.48624440179142675,
+      "grad_norm": 275.15385748139875,
+      "learning_rate": 2.435063518012335e-07,
+      "logits/chosen": -2.592048168182373,
+      "logits/rejected": -2.5161328315734863,
+      "logps/chosen": -257.56207275390625,
+      "logps/rejected": -202.87477111816406,
+      "loss": 0.2628,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.6292481422424316,
+      "rewards/margins": 5.401675224304199,
+      "rewards/rejected": -1.7724273204803467,
       "step": 190
     },
     {
-      "epoch": 0.51,
-      "grad_norm": 859.0725144089504,
-      "learning_rate": 2.823542203635138e-07,
-      "logits/chosen": -2.6617209911346436,
-      "logits/rejected": -2.5823440551757812,
-      "logps/chosen": -278.6440734863281,
-      "logps/rejected": -222.3423309326172,
-      "loss": 0.493,
-      "rewards/accuracies": 0.887499988079071,
-      "rewards/chosen": 1.6911423206329346,
-      "rewards/margins": 9.542765617370605,
-      "rewards/rejected": -7.851624488830566,
+      "epoch": 0.5118362124120281,
+      "grad_norm": 441.7914810102142,
+      "learning_rate": 2.2588337629081105e-07,
+      "logits/chosen": -2.6288845539093018,
+      "logits/rejected": -2.539705753326416,
+      "logps/chosen": -272.2264709472656,
+      "logps/rejected": -217.74020385742188,
+      "loss": 0.2691,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 4.148322105407715,
+      "rewards/margins": 6.209266662597656,
+      "rewards/rejected": -2.0609450340270996,
       "step": 200
     },
     {
-      "epoch": 0.54,
-      "grad_norm": 1073.7778122145994,
-      "learning_rate": 2.600664850273538e-07,
-      "logits/chosen": -2.6473941802978516,
-      "logits/rejected": -2.581458568572998,
-      "logps/chosen": -270.93621826171875,
-      "logps/rejected": -214.9030303955078,
-      "loss": 0.7378,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 0.10583686828613281,
-      "rewards/margins": 7.7938642501831055,
-      "rewards/rejected": -7.688027858734131,
+      "epoch": 0.5374280230326296,
+      "grad_norm": 578.0076877712855,
+      "learning_rate": 2.0805318802188303e-07,
+      "logits/chosen": -2.6119117736816406,
+      "logits/rejected": -2.5380859375,
+      "logps/chosen": -265.3163146972656,
+      "logps/rejected": -211.01132202148438,
+      "loss": 0.4145,
+      "rewards/accuracies": 0.824999988079071,
+      "rewards/chosen": 2.868762969970703,
+      "rewards/margins": 5.194035530090332,
+      "rewards/rejected": -2.32527232170105,
       "step": 210
     },
     {
-      "epoch": 0.56,
-      "grad_norm": 783.9779969138841,
-      "learning_rate": 2.3769816112703045e-07,
-      "logits/chosen": -2.6640350818634033,
-      "logits/rejected": -2.606898307800293,
-      "logps/chosen": -260.6977233886719,
-      "logps/rejected": -216.708740234375,
-      "loss": 0.5298,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 0.881191611289978,
-      "rewards/margins": 7.128912448883057,
-      "rewards/rejected": -6.247720241546631,
+      "epoch": 0.5630198336532309,
+      "grad_norm": 435.59772463261663,
+      "learning_rate": 1.9015852890162436e-07,
+      "logits/chosen": -2.628199577331543,
+      "logits/rejected": -2.5644900798797607,
+      "logps/chosen": -256.01708984375,
+      "logps/rejected": -213.6505126953125,
+      "loss": 0.302,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 2.8298840522766113,
+      "rewards/margins": 4.771709442138672,
+      "rewards/rejected": -1.9418258666992188,
       "step": 220
     },
     {
-      "epoch": 0.59,
-      "grad_norm": 656.1996441950056,
-      "learning_rate": 2.1542832120881677e-07,
-      "logits/chosen": -2.6948089599609375,
-      "logits/rejected": -2.615889072418213,
-      "logps/chosen": -270.77557373046875,
-      "logps/rejected": -219.38320922851562,
-      "loss": 0.4547,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 2.3647713661193848,
-      "rewards/margins": 8.186816215515137,
-      "rewards/rejected": -5.82204532623291,
+      "epoch": 0.5886116442738324,
+      "grad_norm": 341.5067091186977,
+      "learning_rate": 1.723426569670534e-07,
+      "logits/chosen": -2.660271644592285,
+      "logits/rejected": -2.5745387077331543,
+      "logps/chosen": -265.7137145996094,
+      "logps/rejected": -216.05184936523438,
+      "loss": 0.2709,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/chosen": 3.8446857929229736,
+      "rewards/margins": 5.413482666015625,
+      "rewards/rejected": -1.5687971115112305,
       "step": 230
     },
     {
-      "epoch": 0.61,
-      "grad_norm": 1015.3762946599339,
-      "learning_rate": 1.934352493925695e-07,
-      "logits/chosen": -2.6751513481140137,
-      "logits/rejected": -2.6343884468078613,
-      "logps/chosen": -266.14263916015625,
-      "logps/rejected": -223.5273895263672,
-      "loss": 0.429,
-      "rewards/accuracies": 0.8999999761581421,
-      "rewards/chosen": 1.4297797679901123,
-      "rewards/margins": 10.390259742736816,
-      "rewards/rejected": -8.960479736328125,
+      "epoch": 0.6142034548944337,
+      "grad_norm": 518.1474749065682,
+      "learning_rate": 1.547481995140556e-07,
+      "logits/chosen": -2.6460232734680176,
+      "logits/rejected": -2.5979976654052734,
+      "logps/chosen": -260.61566162109375,
+      "logps/rejected": -219.18740844726562,
+      "loss": 0.2732,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.5578017234802246,
+      "rewards/margins": 6.3658447265625,
+      "rewards/rejected": -2.808042287826538,
       "step": 240
     },
     {
-      "epoch": 0.64,
-      "grad_norm": 586.0366917195352,
-      "learning_rate": 1.7189501409486059e-07,
-      "logits/chosen": -2.6815237998962402,
-      "logits/rejected": -2.6186068058013916,
-      "logps/chosen": -271.32818603515625,
-      "logps/rejected": -225.2965545654297,
-      "loss": 0.4243,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 0.4054955542087555,
-      "rewards/margins": 8.429037094116211,
-      "rewards/rejected": -8.023542404174805,
+      "epoch": 0.6397952655150352,
+      "grad_norm": 488.31668996973804,
+      "learning_rate": 1.375160112758885e-07,
+      "logits/chosen": -2.657061815261841,
+      "logits/rejected": -2.5865418910980225,
+      "logps/chosen": -265.78448486328125,
+      "logps/rejected": -221.42001342773438,
+      "loss": 0.2763,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 2.9970998764038086,
+      "rewards/margins": 5.516369819641113,
+      "rewards/rejected": -2.519270420074463,
       "step": 250
     },
     {
-      "epoch": 0.67,
-      "grad_norm": 567.2324686905487,
-      "learning_rate": 1.5098005849021078e-07,
-      "logits/chosen": -2.6706247329711914,
-      "logits/rejected": -2.6183619499206543,
-      "logps/chosen": -265.3566589355469,
-      "logps/rejected": -211.55789184570312,
-      "loss": 0.4026,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 0.452970027923584,
-      "rewards/margins": 7.969753265380859,
-      "rewards/rejected": -7.516783237457275,
+      "epoch": 0.6653870761356366,
+      "grad_norm": 266.72245551152963,
+      "learning_rate": 1.2078404679216862e-07,
+      "logits/chosen": -2.648825168609619,
+      "logits/rejected": -2.589503526687622,
+      "logps/chosen": -260.22369384765625,
+      "logps/rejected": -208.0442352294922,
+      "loss": 0.2613,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 2.818129301071167,
+      "rewards/margins": 5.237301826477051,
+      "rewards/rejected": -2.419172525405884,
       "step": 260
     },
     {
-      "epoch": 0.69,
-      "grad_norm": 651.6922692823155,
-      "learning_rate": 1.30857819994673e-07,
-      "logits/chosen": -2.648026943206787,
-      "logits/rejected": -2.5697438716888428,
-      "logps/chosen": -276.09661865234375,
-      "logps/rejected": -233.24966430664062,
-      "loss": 0.4615,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 0.4385630190372467,
-      "rewards/margins": 11.782003402709961,
-      "rewards/rejected": -11.343441009521484,
+      "epoch": 0.690978886756238,
+      "grad_norm": 410.4072035856202,
+      "learning_rate": 1.0468625599573841e-07,
+      "logits/chosen": -2.6281862258911133,
+      "logits/rejected": -2.5414164066314697,
+      "logps/chosen": -269.1915283203125,
+      "logps/rejected": -228.3280792236328,
+      "loss": 0.3168,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 3.6961987018585205,
+      "rewards/margins": 7.5373077392578125,
+      "rewards/rejected": -3.841109037399292,
       "step": 270
     },
     {
-      "epoch": 0.72,
-      "grad_norm": 492.898374979646,
-      "learning_rate": 1.116893898236716e-07,
-      "logits/chosen": -2.6764636039733887,
-      "logits/rejected": -2.6275370121002197,
-      "logps/chosen": -273.55120849609375,
-      "logps/rejected": -222.28256225585938,
-      "loss": 0.431,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 0.1396121382713318,
-      "rewards/margins": 8.543681144714355,
-      "rewards/rejected": -8.404068946838379,
+      "epoch": 0.7165706973768394,
+      "grad_norm": 230.87089413037836,
+      "learning_rate": 8.935151185893727e-08,
+      "logits/chosen": -2.653529644012451,
+      "logits/rejected": -2.5968880653381348,
+      "logps/chosen": -268.2784729003906,
+      "logps/rejected": -218.708984375,
+      "loss": 0.2585,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 2.713934898376465,
+      "rewards/margins": 5.596070289611816,
+      "rewards/rejected": -2.8821353912353516,
       "step": 280
     },
     {
-      "epoch": 0.74,
-      "grad_norm": 601.4691313440442,
-      "learning_rate": 9.362822335518062e-08,
-      "logits/chosen": -2.639714241027832,
-      "logits/rejected": -2.6006019115448,
-      "logps/chosen": -271.51104736328125,
-      "logps/rejected": -219.78646850585938,
-      "loss": 0.3785,
-      "rewards/accuracies": 0.887499988079071,
-      "rewards/chosen": 0.39154329895973206,
-      "rewards/margins": 8.193072319030762,
-      "rewards/rejected": -7.8015289306640625,
+      "epoch": 0.7421625079974408,
+      "grad_norm": 355.41024702928314,
+      "learning_rate": 7.490257868414449e-08,
+      "logits/chosen": -2.613981246948242,
+      "logits/rejected": -2.5670533180236816,
+      "logps/chosen": -266.4203186035156,
+      "logps/rejected": -216.7047119140625,
+      "loss": 0.2561,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 2.7628657817840576,
+      "rewards/margins": 5.556154727935791,
+      "rewards/rejected": -2.793288469314575,
       "step": 290
     },
     {
-      "epoch": 0.77,
-      "grad_norm": 876.6568683843423,
-      "learning_rate": 7.681891162260015e-08,
-      "logits/chosen": -2.6577625274658203,
-      "logits/rejected": -2.610130786895752,
-      "logps/chosen": -278.3086853027344,
-      "logps/rejected": -223.9765625,
-      "loss": 0.4175,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 0.23941349983215332,
-      "rewards/margins": 8.298933029174805,
-      "rewards/rejected": -8.059518814086914,
+      "epoch": 0.7677543186180422,
+      "grad_norm": 334.3998771127154,
+      "learning_rate": 6.145512929808013e-08,
+      "logits/chosen": -2.6336982250213623,
+      "logits/rejected": -2.579585552215576,
+      "logps/chosen": -273.02044677734375,
+      "logps/rejected": -220.3231964111328,
+      "loss": 0.274,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 2.7771525382995605,
+      "rewards/margins": 5.427972316741943,
+      "rewards/rejected": -2.650820016860962,
       "step": 300
     },
     {
-      "epoch": 0.79,
-      "grad_norm": 491.2362210232499,
-      "learning_rate": 6.139602377230247e-08,
-      "logits/chosen": -2.623286724090576,
-      "logits/rejected": -2.5633485317230225,
-      "logps/chosen": -281.3511047363281,
-      "logps/rejected": -218.5331268310547,
-      "loss": 0.4291,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 0.8305476903915405,
-      "rewards/margins": 8.879867553710938,
-      "rewards/rejected": -8.049318313598633,
+      "epoch": 0.7933461292386437,
+      "grad_norm": 248.8225361148432,
+      "learning_rate": 4.911681901784197e-08,
+      "logits/chosen": -2.5988125801086426,
+      "logits/rejected": -2.5305018424987793,
+      "logps/chosen": -276.5069885253906,
+      "logps/rejected": -215.4237518310547,
+      "loss": 0.3069,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 2.883484125137329,
+      "rewards/margins": 5.80063533782959,
+      "rewards/rejected": -2.917151927947998,
       "step": 310
     },
     {
-      "epoch": 0.82,
-      "grad_norm": 762.3598954431114,
-      "learning_rate": 4.748302975270837e-08,
-      "logits/chosen": -2.6497387886047363,
-      "logits/rejected": -2.610168933868408,
-      "logps/chosen": -264.21588134765625,
-      "logps/rejected": -206.8611602783203,
-      "loss": 0.4476,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 0.6263083219528198,
-      "rewards/margins": 7.853513240814209,
-      "rewards/rejected": -7.227204322814941,
+      "epoch": 0.818937939859245,
+      "grad_norm": 378.21338988259265,
+      "learning_rate": 3.79864238021667e-08,
+      "logits/chosen": -2.621037006378174,
+      "logits/rejected": -2.5753254890441895,
+      "logps/chosen": -259.85833740234375,
+      "logps/rejected": -204.21798706054688,
+      "loss": 0.2691,
+      "rewards/accuracies": 0.8999999761581421,
+      "rewards/chosen": 2.5267233848571777,
+      "rewards/margins": 5.2202558517456055,
+      "rewards/rejected": -2.6935324668884277,
       "step": 320
     },
     {
-      "epoch": 0.84,
-      "grad_norm": 668.8344256472935,
-      "learning_rate": 3.5191311859445795e-08,
-      "logits/chosen": -2.6693060398101807,
-      "logits/rejected": -2.6197760105133057,
-      "logps/chosen": -268.38543701171875,
-      "logps/rejected": -220.74081420898438,
-      "loss": 0.4113,
-      "rewards/accuracies": 0.90625,
-      "rewards/chosen": 1.1819716691970825,
-      "rewards/margins": 8.392292976379395,
-      "rewards/rejected": -7.210320949554443,
+      "epoch": 0.8445297504798465,
+      "grad_norm": 354.9475645655247,
+      "learning_rate": 2.8153049487556634e-08,
+      "logits/chosen": -2.6436188220977783,
+      "logits/rejected": -2.585820436477661,
+      "logps/chosen": -263.09808349609375,
+      "logps/rejected": -217.03341674804688,
+      "loss": 0.2695,
+      "rewards/accuracies": 0.8999999761581421,
+      "rewards/chosen": 3.3003220558166504,
+      "rewards/margins": 5.452363014221191,
+      "rewards/rejected": -2.1520419120788574,
       "step": 330
     },
     {
-      "epoch": 0.87,
-      "grad_norm": 600.9418350776771,
-      "learning_rate": 2.4619273049795996e-08,
-      "logits/chosen": -2.653104066848755,
-      "logits/rejected": -2.6034810543060303,
-      "logps/chosen": -263.5737609863281,
-      "logps/rejected": -213.79544067382812,
-      "loss": 0.3933,
-      "rewards/accuracies": 0.9125000238418579,
-      "rewards/chosen": 1.1136972904205322,
-      "rewards/margins": 9.145872116088867,
-      "rewards/rejected": -8.032175064086914,
+      "epoch": 0.8701215611004478,
+      "grad_norm": 298.79503920742496,
+      "learning_rate": 1.9695418439836796e-08,
+      "logits/chosen": -2.6284804344177246,
+      "logits/rejected": -2.5717878341674805,
+      "logps/chosen": -258.5690002441406,
+      "logps/rejected": -210.4982147216797,
+      "loss": 0.2354,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/chosen": 3.1211013793945312,
+      "rewards/margins": 5.934812545776367,
+      "rewards/rejected": -2.813711643218994,
       "step": 340
     },
     {
-      "epoch": 0.9,
-      "grad_norm": 448.6960620183454,
-      "learning_rate": 1.5851549164932115e-08,
-      "logits/chosen": -2.663357734680176,
-      "logits/rejected": -2.6194939613342285,
-      "logps/chosen": -272.70233154296875,
-      "logps/rejected": -229.30697631835938,
-      "loss": 0.4249,
-      "rewards/accuracies": 0.887499988079071,
-      "rewards/chosen": 1.3936760425567627,
-      "rewards/margins": 8.432661056518555,
-      "rewards/rejected": -7.0389862060546875,
+      "epoch": 0.8957133717210493,
+      "grad_norm": 256.56016530768545,
+      "learning_rate": 1.2681239331945692e-08,
+      "logits/chosen": -2.6408398151397705,
+      "logits/rejected": -2.5925440788269043,
+      "logps/chosen": -268.15728759765625,
+      "logps/rejected": -226.1197967529297,
+      "loss": 0.2666,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.046782970428467,
+      "rewards/margins": 5.363730430603027,
+      "rewards/rejected": -2.3169472217559814,
       "step": 350
     },
     {
-      "epoch": 0.92,
-      "grad_norm": 606.8337189375926,
-      "learning_rate": 8.958331366609423e-09,
-      "logits/chosen": -2.666874885559082,
-      "logits/rejected": -2.6079227924346924,
-      "logps/chosen": -278.23712158203125,
-      "logps/rejected": -222.1784210205078,
-      "loss": 0.4777,
-      "rewards/accuracies": 0.893750011920929,
-      "rewards/chosen": 0.8818317651748657,
-      "rewards/margins": 8.486412048339844,
-      "rewards/rejected": -7.604579925537109,
+      "epoch": 0.9213051823416507,
+      "grad_norm": 316.87066881593194,
+      "learning_rate": 7.166665093287538e-09,
+      "logits/chosen": -2.6406874656677246,
+      "logits/rejected": -2.5731348991394043,
+      "logps/chosen": -273.30157470703125,
+      "logps/rejected": -218.8077392578125,
+      "loss": 0.3057,
+      "rewards/accuracies": 0.90625,
+      "rewards/chosen": 2.9576709270477295,
+      "rewards/margins": 5.49710750579834,
+      "rewards/rejected": -2.539436101913452,
       "step": 360
     },
     {
-      "epoch": 0.95,
-      "grad_norm": 565.1329564450134,
-      "learning_rate": 3.994804212627461e-09,
-      "logits/chosen": -2.626528739929199,
-      "logits/rejected": -2.5960590839385986,
-      "logps/chosen": -277.90789794921875,
-      "logps/rejected": -232.6277313232422,
-      "loss": 0.4164,
-      "rewards/accuracies": 0.84375,
-      "rewards/chosen": 1.226670503616333,
-      "rewards/margins": 8.475992202758789,
-      "rewards/rejected": -7.249321937561035,
+      "epoch": 0.946896992962252,
+      "grad_norm": 569.0448969619218,
+      "learning_rate": 3.1958433701019694e-09,
+      "logits/chosen": -2.6003308296203613,
+      "logits/rejected": -2.5643393993377686,
+      "logps/chosen": -272.14678955078125,
+      "logps/rejected": -228.38943481445312,
+      "loss": 0.3023,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": 3.5620276927948,
+      "rewards/margins": 5.470287322998047,
+      "rewards/rejected": -1.908259391784668,
       "step": 370
     },
     {
-      "epoch": 0.97,
-      "grad_norm": 494.78405010061283,
-      "learning_rate": 1.0007038696262516e-09,
-      "logits/chosen": -2.672760486602783,
-      "logits/rejected": -2.6364920139312744,
-      "logps/chosen": -266.59918212890625,
-      "logps/rejected": -233.4165496826172,
-      "loss": 0.4064,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 1.6075836420059204,
-      "rewards/margins": 8.459333419799805,
-      "rewards/rejected": -6.851749420166016,
+      "epoch": 0.9724888035828535,
+      "grad_norm": 206.4373391747023,
+      "learning_rate": 8.005630957010012e-10,
+      "logits/chosen": -2.6491971015930176,
+      "logits/rejected": -2.6083197593688965,
+      "logps/chosen": -261.6040344238281,
+      "logps/rejected": -229.88302612304688,
+      "loss": 0.273,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.3906753063201904,
+      "rewards/margins": 5.430436611175537,
+      "rewards/rejected": -2.039761543273926,
       "step": 380
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 917.0969156966863,
+      "epoch": 0.9980806142034548,
+      "grad_norm": 529.0802276796763,
       "learning_rate": 0.0,
-      "logits/chosen": -2.6849241256713867,
-      "logits/rejected": -2.630876302719116,
-      "logps/chosen": -254.27664184570312,
-      "logps/rejected": -213.14102172851562,
-      "loss": 0.4983,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 0.6460372805595398,
-      "rewards/margins": 8.144643783569336,
-      "rewards/rejected": -7.498606204986572,
+      "logits/chosen": -2.6596872806549072,
+      "logits/rejected": -2.598179578781128,
+      "logps/chosen": -249.18356323242188,
+      "logps/rejected": -209.78622436523438,
+      "loss": 0.2803,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 2.9054455757141113,
+      "rewards/margins": 5.3939642906188965,
+      "rewards/rejected": -2.488518476486206,
       "step": 390
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.9980806142034548,
       "step": 390,
       "total_flos": 0.0,
-      "train_loss": 0.4513903107398596,
-      "train_runtime": 5854.463,
-      "train_samples_per_second": 8.54,
-      "train_steps_per_second": 0.067
+      "train_loss": 0.3119094243416419,
+      "train_runtime": 6443.086,
+      "train_samples_per_second": 7.76,
+      "train_steps_per_second": 0.061
     }
   ],
   "logging_steps": 10,
@@ -623,6 +623,18 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
   "total_flos": 0.0,
   "train_batch_size": 4,
   "trial_name": null,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa0e00d3aa1ab0a0ef1dd07b8b8bd52c64ef5747a57df47646eee4b22e879a38
-size 6328
+oid sha256:6819c1dd4414cf3e1008e039028d4a09c333504f352311e2138049a4cec39cd4
+size 6520