Model save

Browse files

Files changed (11) hide show

README.md +16 -0
all_results.json +5 -5
config.json +1 -2
model-00001-of-00003.safetensors +1 -1
model-00002-of-00003.safetensors +1 -1
model-00003-of-00003.safetensors +1 -1
tokenizer.json +0 -1
tokenizer_config.json +0 -2
train_results.json +5 -5
trainer_state.json +616 -454
training_args.bin +2 -2

README.md CHANGED Viewed

@@ -16,6 +16,16 @@ should probably proofread and complete it, then remove this comment. -->
 # zephyr-7b-dpo-full
 This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the None dataset.
 ## Model description
@@ -50,6 +60,12 @@ The following hyperparameters were used during training:
 ### Training results
 ### Framework versions

 # zephyr-7b-dpo-full
 This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.0194
+- Rewards/chosen: -1.8199
+- Rewards/rejected: -2.5238
+- Rewards/accuracies: 0.7031
+- Rewards/margins: 0.7039
+- Logps/rejected: -509.7351
+- Logps/chosen: -439.0294
+- Logits/rejected: -2.2378
+- Logits/chosen: -2.2625
 ## Model description
 ### Training results
+| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
+|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
+| 0.0502        | 0.21  | 100  | 0.0391          | -0.7692        | -1.1939          | 0.6914             | 0.4246          | -376.7401      | -333.9633    | -2.7065         | -2.7235       |
+| 0.0237        | 0.42  | 200  | 0.0268          | -1.2426        | -1.8386          | 0.6680             | 0.5960          | -441.2099      | -381.2964    | -2.4965         | -2.5105       |
+| 0.0248        | 0.63  | 300  | 0.0254          | -1.3833        | -2.0353          | 0.6914             | 0.6521          | -460.8870      | -395.3680    | -2.3196         | -2.3397       |
+| 0.0194        | 0.84  | 400  | 0.0194          | -1.8199        | -2.5238          | 0.7031             | 0.7039          | -509.7351      | -439.0294    | -2.2378         | -2.2625       |
 ### Framework versions

all_results.json CHANGED Viewed

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.02901148688002133,
-    "train_runtime": 3226.6015,
-    "train_samples": 51894,
-    "train_samples_per_second": 16.083,
-    "train_steps_per_second": 0.126
 }

 {
     "epoch": 1.0,
+    "train_loss": 0.03868693582495386,
+    "train_runtime": 3950.8818,
+    "train_samples": 61134,
+    "train_samples_per_second": 15.474,
+    "train_steps_per_second": 0.121
 }

config.json CHANGED Viewed

@@ -3,7 +3,6 @@
   "architectures": [
     "MistralForCausalLM"
   ],
-  "attention_dropout": 0.0,
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
@@ -20,7 +19,7 @@
   "sliding_window": 4096,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.41.1",
   "use_cache": false,
   "vocab_size": 32000
 }

   "architectures": [
     "MistralForCausalLM"
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
   "hidden_act": "silu",
   "sliding_window": 4096,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
+  "transformers_version": "4.35.2",
   "use_cache": false,
   "vocab_size": 32000
 }

model-00001-of-00003.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8e23d2fcf85d3633c98976a3baaaab90dbd638a579a4ad13c62c9e5462f3a058
 size 4943162336

 version https://git-lfs.github.com/spec/v1
+oid sha256:472b04e0cc50bda621fe5cbd642762d1767be557c1f585354c28a09172156977
 size 4943162336

model-00002-of-00003.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4cc6f3805f76e68ac4b92a78c11a93c5e6990c6f8d1070bed4d48bf55fb342a2
 size 4999819336

 version https://git-lfs.github.com/spec/v1
+oid sha256:aea925956bfb36d8ce6bf0529378c0a70c0af339b58425c70ea48f9146d63695
 size 4999819336

model-00003-of-00003.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5fb6b66cdd7b6dba191b98993395f84d50be9733154e028f6184263aac5c7c25
 size 4540516344

 version https://git-lfs.github.com/spec/v1
+oid sha256:70949f8d5a33dbbc2818f352d2585dd202e52822e017939aa5d1bc67dd71cac9
 size 4540516344

tokenizer.json CHANGED Viewed

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

     "end_of_word_suffix": null,
     "fuse_unk": true,
     "byte_fallback": true,
     "vocab": {
       "<unk>": 0,
       "<s>": 1,

tokenizer_config.json CHANGED Viewed

@@ -1,6 +1,4 @@
 {
-  "add_bos_token": true,
-  "add_eos_token": false,
   "added_tokens_decoder": {
     "0": {
       "content": "<unk>",

 {
   "added_tokens_decoder": {
     "0": {
       "content": "<unk>",

train_results.json CHANGED Viewed

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.02901148688002133,
-    "train_runtime": 3226.6015,
-    "train_samples": 51894,
-    "train_samples_per_second": 16.083,
-    "train_steps_per_second": 0.126
 }

 {
     "epoch": 1.0,
+    "train_loss": 0.03868693582495386,
+    "train_runtime": 3950.8818,
+    "train_samples": 61134,
+    "train_samples_per_second": 15.474,
+    "train_steps_per_second": 0.121
 }

trainer_state.json CHANGED Viewed

@@ -1,21 +1,21 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.998766954377312,
-  "eval_steps": 1000,
-  "global_step": 405,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 1.2195121951219512e-08,
-      "logits/chosen": -2.8412017822265625,
-      "logits/rejected": -2.798035144805908,
-      "logps/chosen": -481.1488037109375,
-      "logps/rejected": -89.30835723876953,
-      "loss": 0.1683,
       "rewards/accuracies": 0.0,
       "rewards/chosen": 0.0,
       "rewards/margins": 0.0,
@@ -24,576 +24,738 @@
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.219512195121951e-07,
-      "logits/chosen": -2.760549306869507,
-      "logits/rejected": -2.7219324111938477,
-      "logps/chosen": -418.5386962890625,
-      "logps/rejected": -116.76993560791016,
-      "loss": 0.1668,
-      "rewards/accuracies": 0.4791666567325592,
-      "rewards/chosen": 0.0004514391766861081,
-      "rewards/margins": 0.0010180326644331217,
-      "rewards/rejected": -0.0005665934295393527,
       "step": 10
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.439024390243902e-07,
-      "logits/chosen": -2.786688804626465,
-      "logits/rejected": -2.740872859954834,
-      "logps/chosen": -396.0628967285156,
-      "logps/rejected": -109.14559173583984,
-      "loss": 0.1685,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": 0.013597361743450165,
-      "rewards/margins": 0.02673523500561714,
-      "rewards/rejected": -0.013137872330844402,
       "step": 20
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.6585365853658536e-07,
-      "logits/chosen": -2.7145609855651855,
-      "logits/rejected": -2.680011749267578,
-      "logps/chosen": -375.7653503417969,
-      "logps/rejected": -119.87037658691406,
-      "loss": 0.1445,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": 0.05958707258105278,
-      "rewards/margins": 0.15927240252494812,
-      "rewards/rejected": -0.09968532621860504,
       "step": 30
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.878048780487804e-07,
-      "logits/chosen": -2.554194211959839,
-      "logits/rejected": -2.5306203365325928,
-      "logps/chosen": -426.01214599609375,
-      "logps/rejected": -156.78164672851562,
-      "loss": 0.1068,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 0.07754631340503693,
-      "rewards/margins": 0.4439311921596527,
-      "rewards/rejected": -0.3663848638534546,
       "step": 40
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.992461696250783e-07,
-      "logits/chosen": -2.457973003387451,
-      "logits/rejected": -2.458061456680298,
-      "logps/chosen": -359.9261779785156,
-      "logps/rejected": -179.5415802001953,
-      "loss": 0.0678,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.0965609923005104,
-      "rewards/margins": 0.6131138801574707,
-      "rewards/rejected": -0.7096748352050781,
       "step": 50
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.966461721767899e-07,
-      "logits/chosen": -2.4080331325531006,
-      "logits/rejected": -2.3387179374694824,
-      "logps/chosen": -477.7005920410156,
-      "logps/rejected": -213.6736297607422,
-      "loss": 0.048,
-      "rewards/accuracies": 0.8062499761581421,
-      "rewards/chosen": -0.09366194158792496,
-      "rewards/margins": 1.000090479850769,
-      "rewards/rejected": -1.0937525033950806,
       "step": 60
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.922100518015975e-07,
-      "logits/chosen": -2.369874954223633,
-      "logits/rejected": -2.32528018951416,
-      "logps/chosen": -452.83404541015625,
-      "logps/rejected": -244.33273315429688,
-      "loss": 0.0332,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -0.27397626638412476,
-      "rewards/margins": 1.1336454153060913,
-      "rewards/rejected": -1.4076217412948608,
       "step": 70
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.859708325770919e-07,
-      "logits/chosen": -2.4045441150665283,
-      "logits/rejected": -2.361023426055908,
-      "logps/chosen": -457.1015625,
-      "logps/rejected": -260.1266784667969,
-      "loss": 0.0321,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.24543538689613342,
-      "rewards/margins": 1.3190863132476807,
-      "rewards/rejected": -1.5645216703414917,
       "step": 80
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 4.779749614980225e-07,
-      "logits/chosen": -2.383591651916504,
-      "logits/rejected": -2.3290789127349854,
-      "logps/chosen": -447.29949951171875,
-      "logps/rejected": -292.9815368652344,
-      "loss": 0.0229,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -0.4048345983028412,
-      "rewards/margins": 1.4302546977996826,
-      "rewards/rejected": -1.8350893259048462,
       "step": 90
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.682819627081427e-07,
-      "logits/chosen": -2.392648458480835,
-      "logits/rejected": -2.339670181274414,
-      "logps/chosen": -497.0926818847656,
-      "logps/rejected": -332.1764221191406,
-      "loss": 0.0186,
-      "rewards/accuracies": 0.824999988079071,
-      "rewards/chosen": -0.6317045092582703,
-      "rewards/margins": 1.6195662021636963,
-      "rewards/rejected": -2.2512707710266113,
       "step": 100
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.569639943810477e-07,
-      "logits/chosen": -2.383877754211426,
-      "logits/rejected": -2.328397274017334,
-      "logps/chosen": -493.90594482421875,
-      "logps/rejected": -354.2288513183594,
-      "loss": 0.015,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.5901327729225159,
-      "rewards/margins": 1.7880985736846924,
-      "rewards/rejected": -2.3782310485839844,
       "step": 110
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 4.4410531154874543e-07,
-      "logits/chosen": -2.322819232940674,
-      "logits/rejected": -2.2361202239990234,
-      "logps/chosen": -511.83636474609375,
-      "logps/rejected": -377.99505615234375,
-      "loss": 0.014,
-      "rewards/accuracies": 0.8062499761581421,
-      "rewards/chosen": -0.8583305478096008,
-      "rewards/margins": 1.807847261428833,
-      "rewards/rejected": -2.666177749633789,
       "step": 120
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 4.298016388768561e-07,
-      "logits/chosen": -2.2856903076171875,
-      "logits/rejected": -2.205942153930664,
-      "logps/chosen": -532.1744384765625,
-      "logps/rejected": -420.78759765625,
-      "loss": 0.0101,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": -1.339316725730896,
-      "rewards/margins": 1.7573401927947998,
-      "rewards/rejected": -3.0966567993164062,
       "step": 130
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.1415945805573005e-07,
-      "logits/chosen": -2.2981626987457275,
-      "logits/rejected": -2.216963529586792,
-      "logps/chosen": -511.155029296875,
-      "logps/rejected": -384.3242492675781,
-      "loss": 0.0119,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": -0.8295551538467407,
-      "rewards/margins": 1.907530426979065,
-      "rewards/rejected": -2.7370851039886475,
       "step": 140
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.972952151123984e-07,
-      "logits/chosen": -2.299988269805908,
-      "logits/rejected": -2.2516579627990723,
-      "logps/chosen": -495.01983642578125,
-      "logps/rejected": -383.9313659667969,
-      "loss": 0.013,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": -0.8993278741836548,
-      "rewards/margins": 1.656372308731079,
-      "rewards/rejected": -2.5557000637054443,
       "step": 150
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.793344535444142e-07,
-      "logits/chosen": -2.2775027751922607,
-      "logits/rejected": -2.1973774433135986,
-      "logps/chosen": -531.9027709960938,
-      "logps/rejected": -419.78973388671875,
-      "loss": 0.0137,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/chosen": -1.0927542448043823,
-      "rewards/margins": 1.8101270198822021,
-      "rewards/rejected": -2.902881383895874,
       "step": 160
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.604108797288461e-07,
-      "logits/chosen": -2.250196933746338,
-      "logits/rejected": -2.184962749481201,
-      "logps/chosen": -502.00848388671875,
-      "logps/rejected": -357.9796447753906,
-      "loss": 0.0159,
-      "rewards/accuracies": 0.8500000238418579,
-      "rewards/chosen": -0.826766848564148,
-      "rewards/margins": 1.7090247869491577,
-      "rewards/rejected": -2.5357918739318848,
       "step": 170
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.40665367563858e-07,
-      "logits/chosen": -2.2178635597229004,
-      "logits/rejected": -2.142676830291748,
-      "logps/chosen": -526.4637451171875,
-      "logps/rejected": -398.69586181640625,
-      "loss": 0.0125,
-      "rewards/accuracies": 0.8062499761581421,
-      "rewards/chosen": -1.0359699726104736,
-      "rewards/margins": 1.687019944190979,
-      "rewards/rejected": -2.722989559173584,
       "step": 180
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 3.202449097526798e-07,
-      "logits/chosen": -2.2205810546875,
-      "logits/rejected": -2.1265811920166016,
-      "logps/chosen": -558.7676391601562,
-      "logps/rejected": -478.9952087402344,
-      "loss": 0.0106,
-      "rewards/accuracies": 0.8062499761581421,
-      "rewards/chosen": -1.2641270160675049,
-      "rewards/margins": 2.266322612762451,
-      "rewards/rejected": -3.530449628829956,
       "step": 190
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.993015235369905e-07,
-      "logits/chosen": -2.287707805633545,
-      "logits/rejected": -2.2064316272735596,
-      "logps/chosen": -542.5303955078125,
-      "logps/rejected": -418.10614013671875,
-      "loss": 0.0121,
-      "rewards/accuracies": 0.831250011920929,
-      "rewards/chosen": -0.9594089388847351,
-      "rewards/margins": 2.0673024654388428,
-      "rewards/rejected": -3.0267112255096436,
       "step": 200
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 2.7799111902582693e-07,
-      "logits/chosen": -2.2779388427734375,
-      "logits/rejected": -2.217900037765503,
-      "logps/chosen": -507.04736328125,
-      "logps/rejected": -406.27593994140625,
-      "loss": 0.0167,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -0.7974947690963745,
-      "rewards/margins": 1.9174222946166992,
-      "rewards/rejected": -2.7149171829223633,
-      "step": 210
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 2.564723385445869e-07,
-      "logits/chosen": -2.236158847808838,
-      "logits/rejected": -2.1744391918182373,
-      "logps/chosen": -486.0782165527344,
-      "logps/rejected": -399.27587890625,
-      "loss": 0.014,
       "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.9962589144706726,
-      "rewards/margins": 1.6734790802001953,
-      "rewards/rejected": -2.6697378158569336,
       "step": 220
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 2.3490537564442845e-07,
-      "logits/chosen": -2.245281934738159,
-      "logits/rejected": -2.1824960708618164,
-      "logps/chosen": -487.1275329589844,
-      "logps/rejected": -412.4351501464844,
-      "loss": 0.0112,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -1.0274240970611572,
-      "rewards/margins": 1.9516007900238037,
-      "rewards/rejected": -2.97902512550354,
       "step": 230
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 2.1345078256378801e-07,
-      "logits/chosen": -2.2380142211914062,
-      "logits/rejected": -2.1482958793640137,
-      "logps/chosen": -488.2174377441406,
-      "logps/rejected": -404.8780822753906,
-      "loss": 0.0111,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -1.1305696964263916,
-      "rewards/margins": 1.7587543725967407,
-      "rewards/rejected": -2.8893239498138428,
       "step": 240
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.9226827501969865e-07,
-      "logits/chosen": -2.2937111854553223,
-      "logits/rejected": -2.209343433380127,
-      "logps/chosen": -510.5157165527344,
-      "logps/rejected": -376.5196838378906,
-      "loss": 0.0143,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": -0.6733769178390503,
-      "rewards/margins": 1.9475233554840088,
-      "rewards/rejected": -2.6209003925323486,
       "step": 250
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.715155432264775e-07,
-      "logits/chosen": -2.2513487339019775,
-      "logits/rejected": -2.203629732131958,
-      "logps/chosen": -487.1371154785156,
-      "logps/rejected": -376.69622802734375,
-      "loss": 0.0154,
-      "rewards/accuracies": 0.8125,
-      "rewards/chosen": -0.7552510499954224,
-      "rewards/margins": 1.8159189224243164,
-      "rewards/rejected": -2.5711700916290283,
       "step": 260
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.51347077992983e-07,
-      "logits/chosen": -2.2633750438690186,
-      "logits/rejected": -2.1752829551696777,
-      "logps/chosen": -490.79290771484375,
-      "logps/rejected": -407.4229431152344,
-      "loss": 0.0126,
-      "rewards/accuracies": 0.824999988079071,
-      "rewards/chosen": -0.8112813234329224,
-      "rewards/margins": 2.021395444869995,
-      "rewards/rejected": -2.832677125930786,
       "step": 270
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.3191302063739906e-07,
-      "logits/chosen": -2.2145628929138184,
-      "logits/rejected": -2.1269583702087402,
-      "logps/chosen": -496.5582580566406,
-      "logps/rejected": -396.8327941894531,
-      "loss": 0.0118,
-      "rewards/accuracies": 0.831250011920929,
-      "rewards/chosen": -0.9003747701644897,
-      "rewards/margins": 1.9780778884887695,
-      "rewards/rejected": -2.878452777862549,
       "step": 280
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.1335804528119475e-07,
-      "logits/chosen": -2.2652816772460938,
-      "logits/rejected": -2.1710205078125,
-      "logps/chosen": -510.77435302734375,
-      "logps/rejected": -426.341796875,
-      "loss": 0.0122,
-      "rewards/accuracies": 0.8125,
-      "rewards/chosen": -1.0885112285614014,
-      "rewards/margins": 2.0129168033599854,
-      "rewards/rejected": -3.101428508758545,
       "step": 290
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 9.582028184286423e-08,
-      "logits/chosen": -2.234984874725342,
-      "logits/rejected": -2.140819549560547,
-      "logps/chosen": -563.6864013671875,
-      "logps/rejected": -454.06292724609375,
-      "loss": 0.0094,
-      "rewards/accuracies": 0.8125,
-      "rewards/chosen": -1.0196846723556519,
-      "rewards/margins": 2.260946273803711,
-      "rewards/rejected": -3.2806308269500732,
       "step": 300
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 7.943028774907065e-08,
-      "logits/chosen": -2.199270248413086,
-      "logits/rejected": -2.1144025325775146,
-      "logps/chosen": -539.964111328125,
-      "logps/rejected": -449.0367736816406,
-      "loss": 0.0102,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -1.2019588947296143,
-      "rewards/margins": 2.055799722671509,
-      "rewards/rejected": -3.2577590942382812,
       "step": 310
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 6.431007601814637e-08,
-      "logits/chosen": -2.2333831787109375,
-      "logits/rejected": -2.1530561447143555,
-      "logps/chosen": -524.09033203125,
-      "logps/rejected": -443.92449951171875,
-      "loss": 0.0109,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -1.2250614166259766,
-      "rewards/margins": 2.0103707313537598,
-      "rewards/rejected": -3.2354321479797363,
       "step": 320
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 5.0572206951246e-08,
-      "logits/chosen": -2.2508292198181152,
-      "logits/rejected": -2.168816328048706,
-      "logps/chosen": -533.7239990234375,
-      "logps/rejected": -462.59722900390625,
-      "loss": 0.0133,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -1.1595772504806519,
-      "rewards/margins": 2.1093368530273438,
-      "rewards/rejected": -3.268913984298706,
       "step": 330
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 3.831895019292897e-08,
-      "logits/chosen": -2.1972341537475586,
-      "logits/rejected": -2.0935590267181396,
-      "logps/chosen": -499.79541015625,
-      "logps/rejected": -410.91278076171875,
-      "loss": 0.01,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/chosen": -1.1320897340774536,
-      "rewards/margins": 1.9490875005722046,
-      "rewards/rejected": -3.0811774730682373,
       "step": 340
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.764152339909756e-08,
-      "logits/chosen": -2.2604408264160156,
-      "logits/rejected": -2.1542410850524902,
-      "logps/chosen": -529.6370239257812,
-      "logps/rejected": -436.3409118652344,
-      "loss": 0.0094,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -1.1161201000213623,
-      "rewards/margins": 2.068046808242798,
-      "rewards/rejected": -3.184166431427002,
       "step": 350
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.861941317991664e-08,
-      "logits/chosen": -2.2331619262695312,
-      "logits/rejected": -2.1493821144104004,
-      "logps/chosen": -487.66253662109375,
-      "logps/rejected": -430.4197692871094,
-      "loss": 0.0113,
-      "rewards/accuracies": 0.8125,
-      "rewards/chosen": -1.1774896383285522,
-      "rewards/margins": 2.0229172706604004,
-      "rewards/rejected": -3.2004075050354004,
       "step": 360
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.13197833728636e-08,
-      "logits/chosen": -2.2261927127838135,
-      "logits/rejected": -2.1371588706970215,
-      "logps/chosen": -520.0165405273438,
-      "logps/rejected": -424.5750427246094,
-      "loss": 0.01,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -0.9708250164985657,
-      "rewards/margins": 2.168119430541992,
-      "rewards/rejected": -3.138944149017334,
       "step": 370
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 5.79697505093521e-09,
-      "logits/chosen": -2.154387950897217,
-      "logits/rejected": -2.0763607025146484,
-      "logps/chosen": -512.8585205078125,
-      "logps/rejected": -427.20465087890625,
-      "loss": 0.0112,
-      "rewards/accuracies": 0.8125,
-      "rewards/chosen": -1.175691843032837,
-      "rewards/margins": 2.0045082569122314,
-      "rewards/rejected": -3.1802000999450684,
       "step": 380
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.092101988131256e-09,
-      "logits/chosen": -2.2704200744628906,
-      "logits/rejected": -2.173677444458008,
-      "logps/chosen": -558.5034790039062,
-      "logps/rejected": -441.8016052246094,
-      "loss": 0.0101,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -1.062382698059082,
-      "rewards/margins": 2.1931846141815186,
-      "rewards/rejected": -3.2555670738220215,
       "step": 390
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 2.327445937151673e-10,
-      "logits/chosen": -2.2109153270721436,
-      "logits/rejected": -2.132047176361084,
-      "logps/chosen": -552.7464599609375,
-      "logps/rejected": -436.6908264160156,
-      "loss": 0.0093,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -1.1525399684906006,
-      "rewards/margins": 2.0077481269836426,
-      "rewards/rejected": -3.160288095474243,
       "step": 400
     },
     {
       "epoch": 1.0,
-      "step": 405,
       "total_flos": 0.0,
-      "train_loss": 0.02901148688002133,
-      "train_runtime": 3226.6015,
-      "train_samples_per_second": 16.083,
-      "train_steps_per_second": 0.126
     }
   ],
   "logging_steps": 10,
-  "max_steps": 405,
   "num_train_epochs": 1,
   "save_steps": 1000,
   "total_flos": 0.0,

 {
   "best_metric": null,
   "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 100,
+  "global_step": 478,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
+      "learning_rate": 1.0416666666666666e-08,
+      "logits/chosen": -2.7050857543945312,
+      "logits/rejected": -2.7461352348327637,
+      "logps/chosen": -137.25845336914062,
+      "logps/rejected": -163.38693237304688,
+      "loss": 0.0968,
       "rewards/accuracies": 0.0,
       "rewards/chosen": 0.0,
       "rewards/margins": 0.0,
     },
     {
       "epoch": 0.02,
+      "learning_rate": 1.0416666666666667e-07,
+      "logits/chosen": -2.731086015701294,
+      "logits/rejected": -2.69549298286438,
+      "logps/chosen": -255.12014770507812,
+      "logps/rejected": -238.85885620117188,
+      "loss": 0.1059,
+      "rewards/accuracies": 0.4722222089767456,
+      "rewards/chosen": 0.00021500665752682835,
+      "rewards/margins": 0.00010767025378299877,
+      "rewards/rejected": 0.00010733638919191435,
       "step": 10
     },
     {
+      "epoch": 0.04,
+      "learning_rate": 2.0833333333333333e-07,
+      "logits/chosen": -2.7516608238220215,
+      "logits/rejected": -2.7480316162109375,
+      "logps/chosen": -265.412109375,
+      "logps/rejected": -255.61865234375,
+      "loss": 0.1048,
+      "rewards/accuracies": 0.5687500238418579,
+      "rewards/chosen": -4.060972059960477e-05,
+      "rewards/margins": 0.0010573023464530706,
+      "rewards/rejected": -0.0010979120852425694,
       "step": 20
     },
     {
+      "epoch": 0.06,
+      "learning_rate": 3.1249999999999997e-07,
+      "logits/chosen": -2.741386651992798,
+      "logits/rejected": -2.698482036590576,
+      "logps/chosen": -253.40029907226562,
+      "logps/rejected": -252.72500610351562,
+      "loss": 0.1012,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.0009587205713614821,
+      "rewards/margins": 0.009668193757534027,
+      "rewards/rejected": -0.010626914910972118,
       "step": 30
     },
     {
+      "epoch": 0.08,
+      "learning_rate": 4.1666666666666667e-07,
+      "logits/chosen": -2.792792797088623,
+      "logits/rejected": -2.778289556503296,
+      "logps/chosen": -274.712158203125,
+      "logps/rejected": -279.0121154785156,
+      "loss": 0.1022,
+      "rewards/accuracies": 0.612500011920929,
+      "rewards/chosen": -0.0003971463011112064,
+      "rewards/margins": 0.02575758472084999,
+      "rewards/rejected": -0.02615473046898842,
       "step": 40
     },
     {
+      "epoch": 0.1,
+      "learning_rate": 4.999733114418725e-07,
+      "logits/chosen": -2.760355234146118,
+      "logits/rejected": -2.7391133308410645,
+      "logps/chosen": -286.1962890625,
+      "logps/rejected": -295.90069580078125,
+      "loss": 0.0957,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -0.041570521891117096,
+      "rewards/margins": 0.03570377081632614,
+      "rewards/rejected": -0.07727430015802383,
       "step": 50
     },
     {
+      "epoch": 0.13,
+      "learning_rate": 4.990398100856366e-07,
+      "logits/chosen": -2.776111125946045,
+      "logits/rejected": -2.7512311935424805,
+      "logps/chosen": -276.09942626953125,
+      "logps/rejected": -258.3904724121094,
+      "loss": 0.0889,
+      "rewards/accuracies": 0.6187499761581421,
+      "rewards/chosen": -0.06735710799694061,
+      "rewards/margins": 0.1039455384016037,
+      "rewards/rejected": -0.1713026612997055,
       "step": 60
     },
     {
+      "epoch": 0.15,
+      "learning_rate": 4.967775735898179e-07,
+      "logits/chosen": -2.768890857696533,
+      "logits/rejected": -2.7490429878234863,
+      "logps/chosen": -270.66156005859375,
+      "logps/rejected": -301.940185546875,
+      "loss": 0.0783,
+      "rewards/accuracies": 0.612500011920929,
+      "rewards/chosen": -0.15578190982341766,
+      "rewards/margins": 0.11446709930896759,
+      "rewards/rejected": -0.27024897933006287,
       "step": 70
     },
     {
+      "epoch": 0.17,
+      "learning_rate": 4.931986719649298e-07,
+      "logits/chosen": -2.7411890029907227,
+      "logits/rejected": -2.71295428276062,
+      "logps/chosen": -270.99969482421875,
+      "logps/rejected": -290.6012878417969,
+      "loss": 0.0734,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": -0.17659564316272736,
+      "rewards/margins": 0.2667813003063202,
+      "rewards/rejected": -0.44337692856788635,
       "step": 80
     },
     {
+      "epoch": 0.19,
+      "learning_rate": 4.883222001996351e-07,
+      "logits/chosen": -2.7742576599121094,
+      "logits/rejected": -2.7353005409240723,
+      "logps/chosen": -315.1212463378906,
+      "logps/rejected": -318.26153564453125,
+      "loss": 0.0628,
+      "rewards/accuracies": 0.7250000238418579,
+      "rewards/chosen": -0.37408870458602905,
+      "rewards/margins": 0.32524389028549194,
+      "rewards/rejected": -0.699332594871521,
       "step": 90
     },
     {
+      "epoch": 0.21,
+      "learning_rate": 4.821741763807186e-07,
+      "logits/chosen": -2.7804133892059326,
+      "logits/rejected": -2.755944013595581,
+      "logps/chosen": -322.177978515625,
+      "logps/rejected": -350.9091796875,
+      "loss": 0.0502,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -0.5584506988525391,
+      "rewards/margins": 0.32031145691871643,
+      "rewards/rejected": -0.8787621259689331,
       "step": 100
     },
     {
+      "epoch": 0.21,
+      "eval_logits/chosen": -2.7234888076782227,
+      "eval_logits/rejected": -2.7065277099609375,
+      "eval_logps/chosen": -333.9632873535156,
+      "eval_logps/rejected": -376.7401428222656,
+      "eval_loss": 0.03910641372203827,
+      "eval_rewards/accuracies": 0.69140625,
+      "eval_rewards/chosen": -0.76923668384552,
+      "eval_rewards/margins": 0.4246319532394409,
+      "eval_rewards/rejected": -1.193868637084961,
+      "eval_runtime": 53.787,
+      "eval_samples_per_second": 37.184,
+      "eval_steps_per_second": 0.595,
+      "step": 100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.747874028753375e-07,
+      "logits/chosen": -2.6888976097106934,
+      "logits/rejected": -2.650428056716919,
+      "logps/chosen": -366.8619079589844,
+      "logps/rejected": -376.2767028808594,
+      "loss": 0.0371,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": -0.7290430665016174,
+      "rewards/margins": 0.3944670259952545,
+      "rewards/rejected": -1.1235100030899048,
       "step": 110
     },
     {
+      "epoch": 0.25,
+      "learning_rate": 4.662012913161997e-07,
+      "logits/chosen": -2.733076810836792,
+      "logits/rejected": -2.6883604526519775,
+      "logps/chosen": -342.3517150878906,
+      "logps/rejected": -385.7752990722656,
+      "loss": 0.0398,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -0.7808613181114197,
+      "rewards/margins": 0.501408040523529,
+      "rewards/rejected": -1.2822694778442383,
       "step": 120
     },
     {
+      "epoch": 0.27,
+      "learning_rate": 4.5646165232345103e-07,
+      "logits/chosen": -2.709172248840332,
+      "logits/rejected": -2.6797726154327393,
+      "logps/chosen": -348.2884216308594,
+      "logps/rejected": -364.2542724609375,
+      "loss": 0.0333,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -0.8973625898361206,
+      "rewards/margins": 0.41800689697265625,
+      "rewards/rejected": -1.3153694868087769,
       "step": 130
     },
     {
+      "epoch": 0.29,
+      "learning_rate": 4.456204510851956e-07,
+      "logits/chosen": -2.6971547603607178,
+      "logits/rejected": -2.6667492389678955,
+      "logps/chosen": -369.86016845703125,
+      "logps/rejected": -398.45721435546875,
+      "loss": 0.0363,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -0.8229220509529114,
+      "rewards/margins": 0.38307899236679077,
+      "rewards/rejected": -1.2060010433197021,
       "step": 140
     },
     {
+      "epoch": 0.31,
+      "learning_rate": 4.337355301007335e-07,
+      "logits/chosen": -2.5674502849578857,
+      "logits/rejected": -2.5130200386047363,
+      "logps/chosen": -363.6653747558594,
+      "logps/rejected": -388.075439453125,
+      "loss": 0.0322,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -0.874055027961731,
+      "rewards/margins": 0.542679488658905,
+      "rewards/rejected": -1.4167345762252808,
       "step": 150
     },
     {
+      "epoch": 0.33,
+      "learning_rate": 4.2087030056579986e-07,
+      "logits/chosen": -2.5751426219940186,
+      "logits/rejected": -2.5513062477111816,
+      "logps/chosen": -367.0979919433594,
+      "logps/rejected": -382.1893615722656,
+      "loss": 0.0365,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -0.9746230840682983,
+      "rewards/margins": 0.31770768761634827,
+      "rewards/rejected": -1.2923307418823242,
       "step": 160
     },
     {
+      "epoch": 0.36,
+      "learning_rate": 4.070934040463998e-07,
+      "logits/chosen": -2.5913219451904297,
+      "logits/rejected": -2.5757110118865967,
+      "logps/chosen": -371.7869567871094,
+      "logps/rejected": -361.24609375,
+      "loss": 0.0393,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.6607332825660706,
+      "rewards/margins": 0.31646376848220825,
+      "rewards/rejected": -0.977196991443634,
       "step": 170
     },
     {
+      "epoch": 0.38,
+      "learning_rate": 3.9247834624635404e-07,
+      "logits/chosen": -2.561448812484741,
+      "logits/rejected": -2.5715978145599365,
+      "logps/chosen": -352.5904846191406,
+      "logps/rejected": -386.1707763671875,
+      "loss": 0.0355,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -0.8926581144332886,
+      "rewards/margins": 0.39244139194488525,
+      "rewards/rejected": -1.2850996255874634,
       "step": 180
     },
     {
+      "epoch": 0.4,
+      "learning_rate": 3.7710310482256523e-07,
+      "logits/chosen": -2.510988473892212,
+      "logits/rejected": -2.493633985519409,
+      "logps/chosen": -343.94232177734375,
+      "logps/rejected": -387.73297119140625,
+      "loss": 0.0284,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.0100765228271484,
+      "rewards/margins": 0.5000672340393066,
+      "rewards/rejected": -1.5101438760757446,
       "step": 190
     },
     {
+      "epoch": 0.42,
+      "learning_rate": 3.610497133404795e-07,
+      "logits/chosen": -2.456496238708496,
+      "logits/rejected": -2.442700147628784,
+      "logps/chosen": -379.01141357421875,
+      "logps/rejected": -432.1864318847656,
+      "loss": 0.0237,
+      "rewards/accuracies": 0.643750011920929,
+      "rewards/chosen": -1.3147755861282349,
+      "rewards/margins": 0.4870442748069763,
+      "rewards/rejected": -1.8018200397491455,
       "step": 200
     },
     {
+      "epoch": 0.42,
+      "eval_logits/chosen": -2.5105109214782715,
+      "eval_logits/rejected": -2.496497392654419,
+      "eval_logps/chosen": -381.2964172363281,
+      "eval_logps/rejected": -441.2099304199219,
+      "eval_loss": 0.026758365333080292,
+      "eval_rewards/accuracies": 0.66796875,
+      "eval_rewards/chosen": -1.242567539215088,
+      "eval_rewards/margins": 0.5959988832473755,
+      "eval_rewards/rejected": -1.8385663032531738,
+      "eval_runtime": 53.7251,
+      "eval_samples_per_second": 37.227,
+      "eval_steps_per_second": 0.596,
+      "step": 200
     },
     {
+      "epoch": 0.44,
+      "learning_rate": 3.4440382358952115e-07,
+      "logits/chosen": -2.464259624481201,
+      "logits/rejected": -2.4774632453918457,
+      "logps/chosen": -408.1732482910156,
+      "logps/rejected": -461.95947265625,
+      "loss": 0.0259,
       "rewards/accuracies": 0.768750011920929,
+      "rewards/chosen": -1.0447111129760742,
+      "rewards/margins": 0.6814562678337097,
+      "rewards/rejected": -1.7261674404144287,
+      "step": 210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.272542485937368e-07,
+      "logits/chosen": -2.454061985015869,
+      "logits/rejected": -2.4306459426879883,
+      "logps/chosen": -375.48541259765625,
+      "logps/rejected": -397.7698059082031,
+      "loss": 0.0304,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -1.1511690616607666,
+      "rewards/margins": 0.46213898062705994,
+      "rewards/rejected": -1.6133081912994385,
       "step": 220
     },
     {
+      "epoch": 0.48,
+      "learning_rate": 3.096924887558854e-07,
+      "logits/chosen": -2.4601683616638184,
+      "logits/rejected": -2.445401430130005,
+      "logps/chosen": -377.16510009765625,
+      "logps/rejected": -408.50640869140625,
+      "loss": 0.0273,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -1.183936595916748,
+      "rewards/margins": 0.452188640832901,
+      "rewards/rejected": -1.6361253261566162,
       "step": 230
     },
     {
+      "epoch": 0.5,
+      "learning_rate": 2.9181224366319943e-07,
+      "logits/chosen": -2.386798620223999,
+      "logits/rejected": -2.3921284675598145,
+      "logps/chosen": -433.074951171875,
+      "logps/rejected": -453.2076110839844,
+      "loss": 0.0283,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": -1.21742844581604,
+      "rewards/margins": 0.5803453326225281,
+      "rewards/rejected": -1.7977739572525024,
       "step": 240
     },
     {
+      "epoch": 0.52,
+      "learning_rate": 2.7370891215954565e-07,
+      "logits/chosen": -2.4028992652893066,
+      "logits/rejected": -2.392509937286377,
+      "logps/chosen": -390.3121032714844,
+      "logps/rejected": -434.3692321777344,
+      "loss": 0.0275,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": -1.207838535308838,
+      "rewards/margins": 0.5656585097312927,
+      "rewards/rejected": -1.773497223854065,
       "step": 250
     },
     {
+      "epoch": 0.54,
+      "learning_rate": 2.55479083351317e-07,
+      "logits/chosen": -2.3397555351257324,
+      "logits/rejected": -2.3070874214172363,
+      "logps/chosen": -451.30047607421875,
+      "logps/rejected": -458.3089294433594,
+      "loss": 0.0239,
+      "rewards/accuracies": 0.731249988079071,
+      "rewards/chosen": -1.6053054332733154,
+      "rewards/margins": 0.562503457069397,
+      "rewards/rejected": -2.167808771133423,
       "step": 260
     },
     {
+      "epoch": 0.56,
+      "learning_rate": 2.3722002126275822e-07,
+      "logits/chosen": -2.328000068664551,
+      "logits/rejected": -2.298135280609131,
+      "logps/chosen": -414.98504638671875,
+      "logps/rejected": -454.4877014160156,
+      "loss": 0.0259,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -1.4466311931610107,
+      "rewards/margins": 0.555857241153717,
+      "rewards/rejected": -2.002488374710083,
       "step": 270
     },
     {
+      "epoch": 0.59,
+      "learning_rate": 2.19029145890313e-07,
+      "logits/chosen": -2.3714160919189453,
+      "logits/rejected": -2.3525238037109375,
+      "logps/chosen": -457.384033203125,
+      "logps/rejected": -497.83660888671875,
+      "loss": 0.0262,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.3860464096069336,
+      "rewards/margins": 0.6043224930763245,
+      "rewards/rejected": -1.9903688430786133,
       "step": 280
     },
     {
+      "epoch": 0.61,
+      "learning_rate": 2.0100351342479216e-07,
+      "logits/chosen": -2.3231723308563232,
+      "logits/rejected": -2.3119874000549316,
+      "logps/chosen": -416.6573791503906,
+      "logps/rejected": -427.49945068359375,
+      "loss": 0.0251,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -1.4526689052581787,
+      "rewards/margins": 0.5181004405021667,
+      "rewards/rejected": -1.9707695245742798,
       "step": 290
     },
     {
+      "epoch": 0.63,
+      "learning_rate": 1.8323929841460178e-07,
+      "logits/chosen": -2.3192086219787598,
+      "logits/rejected": -2.2759287357330322,
+      "logps/chosen": -440.58935546875,
+      "logps/rejected": -448.8992614746094,
+      "loss": 0.0248,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -1.4383071660995483,
+      "rewards/margins": 0.5324681997299194,
+      "rewards/rejected": -1.9707752466201782,
+      "step": 300
+    },
+    {
+      "epoch": 0.63,
+      "eval_logits/chosen": -2.339747190475464,
+      "eval_logits/rejected": -2.319640874862671,
+      "eval_logps/chosen": -395.36798095703125,
+      "eval_logps/rejected": -460.886962890625,
+      "eval_loss": 0.025441216304898262,
+      "eval_rewards/accuracies": 0.69140625,
+      "eval_rewards/chosen": -1.383283257484436,
+      "eval_rewards/margins": 0.6520535945892334,
+      "eval_rewards/rejected": -2.03533673286438,
+      "eval_runtime": 53.7578,
+      "eval_samples_per_second": 37.204,
+      "eval_steps_per_second": 0.595,
       "step": 300
     },
     {
+      "epoch": 0.65,
+      "learning_rate": 1.6583128063291573e-07,
+      "logits/chosen": -2.256671667098999,
+      "logits/rejected": -2.2072629928588867,
+      "logps/chosen": -441.985107421875,
+      "logps/rejected": -454.2076721191406,
+      "loss": 0.0255,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": -1.5002824068069458,
+      "rewards/margins": 0.5682298541069031,
+      "rewards/rejected": -2.068512439727783,
       "step": 310
     },
     {
+      "epoch": 0.67,
+      "learning_rate": 1.488723393865766e-07,
+      "logits/chosen": -2.318877696990967,
+      "logits/rejected": -2.311671495437622,
+      "logps/chosen": -368.1661071777344,
+      "logps/rejected": -419.93499755859375,
+      "loss": 0.0257,
+      "rewards/accuracies": 0.731249988079071,
+      "rewards/chosen": -1.1376068592071533,
+      "rewards/margins": 0.6408455967903137,
+      "rewards/rejected": -1.7784525156021118,
       "step": 320
     },
     {
+      "epoch": 0.69,
+      "learning_rate": 1.3245295796480788e-07,
+      "logits/chosen": -2.307875156402588,
+      "logits/rejected": -2.287741184234619,
+      "logps/chosen": -388.35369873046875,
+      "logps/rejected": -418.2744140625,
+      "loss": 0.0264,
+      "rewards/accuracies": 0.6312500238418579,
+      "rewards/chosen": -1.2828859090805054,
+      "rewards/margins": 0.3729013502597809,
+      "rewards/rejected": -1.6557872295379639,
       "step": 330
     },
     {
+      "epoch": 0.71,
+      "learning_rate": 1.1666074087171627e-07,
+      "logits/chosen": -2.3257226943969727,
+      "logits/rejected": -2.3047900199890137,
+      "logps/chosen": -400.9947204589844,
+      "logps/rejected": -466.18438720703125,
+      "loss": 0.0244,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": -1.4033421277999878,
+      "rewards/margins": 0.6501402854919434,
+      "rewards/rejected": -2.0534825325012207,
       "step": 340
     },
     {
+      "epoch": 0.73,
+      "learning_rate": 1.0157994641835734e-07,
+      "logits/chosen": -2.2791500091552734,
+      "logits/rejected": -2.228987216949463,
+      "logps/chosen": -438.2213439941406,
+      "logps/rejected": -482.71185302734375,
+      "loss": 0.0231,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -1.380692720413208,
+      "rewards/margins": 0.6693974733352661,
+      "rewards/rejected": -2.0500900745391846,
       "step": 350
     },
     {
+      "epoch": 0.75,
+      "learning_rate": 8.729103716819111e-08,
+      "logits/chosen": -2.3203213214874268,
+      "logits/rejected": -2.2689433097839355,
+      "logps/chosen": -447.54644775390625,
+      "logps/rejected": -481.5086364746094,
+      "loss": 0.0244,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -1.4591137170791626,
+      "rewards/margins": 0.7254992723464966,
+      "rewards/rejected": -2.184612989425659,
       "step": 360
     },
     {
+      "epoch": 0.77,
+      "learning_rate": 7.387025063449081e-08,
+      "logits/chosen": -2.2618608474731445,
+      "logits/rejected": -2.2370240688323975,
+      "logps/chosen": -463.3082580566406,
+      "logps/rejected": -520.4660034179688,
+      "loss": 0.0211,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.6034772396087646,
+      "rewards/margins": 0.8156415820121765,
+      "rewards/rejected": -2.419118881225586,
       "step": 370
     },
     {
+      "epoch": 0.79,
+      "learning_rate": 6.138919252022435e-08,
+      "logits/chosen": -2.198676347732544,
+      "logits/rejected": -2.1428768634796143,
+      "logps/chosen": -469.5831604003906,
+      "logps/rejected": -499.9231872558594,
+      "loss": 0.0227,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -1.7702577114105225,
+      "rewards/margins": 0.7163289189338684,
+      "rewards/rejected": -2.4865870475769043,
       "step": 380
     },
     {
+      "epoch": 0.82,
+      "learning_rate": 4.991445467064689e-08,
+      "logits/chosen": -2.2201147079467773,
+      "logits/rejected": -2.207123041152954,
+      "logps/chosen": -446.3898010253906,
+      "logps/rejected": -482.3876953125,
+      "loss": 0.0202,
+      "rewards/accuracies": 0.65625,
+      "rewards/chosen": -1.8155654668807983,
+      "rewards/margins": 0.5150431394577026,
+      "rewards/rejected": -2.330608367919922,
       "step": 390
     },
     {
+      "epoch": 0.84,
+      "learning_rate": 3.9507259776993954e-08,
+      "logits/chosen": -2.2568085193634033,
+      "logits/rejected": -2.2447478771209717,
+      "logps/chosen": -449.7113342285156,
+      "logps/rejected": -495.636474609375,
+      "loss": 0.0194,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -1.930235505104065,
+      "rewards/margins": 0.5124340653419495,
+      "rewards/rejected": -2.442669630050659,
+      "step": 400
+    },
+    {
+      "epoch": 0.84,
+      "eval_logits/chosen": -2.262547254562378,
+      "eval_logits/rejected": -2.237802267074585,
+      "eval_logps/chosen": -439.0294189453125,
+      "eval_logps/rejected": -509.7351379394531,
+      "eval_loss": 0.019433461129665375,
+      "eval_rewards/accuracies": 0.703125,
+      "eval_rewards/chosen": -1.8198977708816528,
+      "eval_rewards/margins": 0.7039206027984619,
+      "eval_rewards/rejected": -2.523818254470825,
+      "eval_runtime": 53.7532,
+      "eval_samples_per_second": 37.207,
+      "eval_steps_per_second": 0.595,
       "step": 400
     },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.022313472693447e-08,
+      "logits/chosen": -2.219531536102295,
+      "logits/rejected": -2.177171468734741,
+      "logps/chosen": -435.68316650390625,
+      "logps/rejected": -480.1519470214844,
+      "loss": 0.0193,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -1.8031377792358398,
+      "rewards/margins": 0.5790697336196899,
+      "rewards/rejected": -2.3822076320648193,
+      "step": 410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2111614344599684e-08,
+      "logits/chosen": -2.261929988861084,
+      "logits/rejected": -2.20975661277771,
+      "logps/chosen": -446.5043029785156,
+      "logps/rejected": -468.5169982910156,
+      "loss": 0.0187,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -1.7549480199813843,
+      "rewards/margins": 0.5446823239326477,
+      "rewards/rejected": -2.2996301651000977,
+      "step": 420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.521597710086439e-08,
+      "logits/chosen": -2.2168891429901123,
+      "logits/rejected": -2.1971139907836914,
+      "logps/chosen": -465.6109313964844,
+      "logps/rejected": -488.16912841796875,
+      "loss": 0.0192,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": -1.8082774877548218,
+      "rewards/margins": 0.5722709894180298,
+      "rewards/rejected": -2.3805484771728516,
+      "step": 430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.57301420397924e-09,
+      "logits/chosen": -2.231231212615967,
+      "logits/rejected": -2.209394931793213,
+      "logps/chosen": -448.60986328125,
+      "logps/rejected": -509.45770263671875,
+      "loss": 0.0187,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -1.8875834941864014,
+      "rewards/margins": 0.6209200620651245,
+      "rewards/rejected": -2.5085036754608154,
+      "step": 440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.212833302556258e-09,
+      "logits/chosen": -2.2720608711242676,
+      "logits/rejected": -2.2489547729492188,
+      "logps/chosen": -436.8863830566406,
+      "logps/rejected": -479.250244140625,
+      "loss": 0.0196,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -1.7134939432144165,
+      "rewards/margins": 0.5517303347587585,
+      "rewards/rejected": -2.2652244567871094,
+      "step": 450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.158697848236607e-09,
+      "logits/chosen": -2.2524914741516113,
+      "logits/rejected": -2.1975879669189453,
+      "logps/chosen": -454.9847717285156,
+      "logps/rejected": -471.0829162597656,
+      "loss": 0.0196,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -1.871111273765564,
+      "rewards/margins": 0.5668479204177856,
+      "rewards/rejected": -2.4379591941833496,
+      "step": 460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.269029751107489e-10,
+      "logits/chosen": -2.2418127059936523,
+      "logits/rejected": -2.2091026306152344,
+      "logps/chosen": -438.8106384277344,
+      "logps/rejected": -487.1082458496094,
+      "loss": 0.0201,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.6759856939315796,
+      "rewards/margins": 0.6907237768173218,
+      "rewards/rejected": -2.3667094707489014,
+      "step": 470
+    },
     {
       "epoch": 1.0,
+      "step": 478,
       "total_flos": 0.0,
+      "train_loss": 0.03868693582495386,
+      "train_runtime": 3950.8818,
+      "train_samples_per_second": 15.474,
+      "train_steps_per_second": 0.121
     }
   ],
   "logging_steps": 10,
+  "max_steps": 478,
   "num_train_epochs": 1,
   "save_steps": 1000,
   "total_flos": 0.0,

training_args.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19e290310c5cdc37c4c14a531cf89528ab0416cf966910b48c267456558d7457
-size 6456

 version https://git-lfs.github.com/spec/v1
+oid sha256:aa0d4dc1074b63bfb44f5057f5f2e0432b683d2eeab8a75dfa36408b5eb7136d
+size 5944