Model save

README.md

@@ -0,0 +1,75 @@
+---
+license: mit
+library_name: peft
+tags:
+- trl
+- dpo
+- generated_from_trainer
+base_model: microsoft/phi-2
+model-index:
+- name: phi-2-gpo-renew2-b0.001-v4-i1
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# phi-2-gpo-renew2-b0.001-v4-i1
+
+This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.0536
+- Rewards/chosen: -0.0036
+- Rewards/rejected: -0.0039
+- Rewards/accuracies: 0.4695
+- Rewards/margins: 0.0002
+- Logps/rejected: -371.0876
+- Logps/chosen: -399.9150
+- Logits/rejected: -0.7623
+- Logits/chosen: -0.8574
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-06
+- train_batch_size: 4
+- eval_batch_size: 4
+- seed: 42
+- distributed_type: multi-GPU
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 16
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 1
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
+|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
+| 0.1203        | 0.32  | 100  | 0.0537          | -0.0024        | -0.0024          | 0.4555             | 0.0001          | -369.6694      | -398.6797    | -0.7167         | -0.8167       |
+| 0.1671        | 0.64  | 200  | 0.0537          | -0.0036        | -0.0037          | 0.4670             | 0.0001          | -370.9240      | -399.8586    | -0.7745         | -0.8674       |
+| 0.1393        | 0.96  | 300  | 0.0536          | -0.0038        | -0.0040          | 0.4625             | 0.0003          | -371.2791      | -400.0731    | -0.7820         | -0.8772       |
+
+
+### Framework versions
+
+- PEFT 0.7.1
+- Transformers 4.36.2
+- Pytorch 2.1.2
+- Datasets 2.14.6
+- Tokenizers 0.15.2

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92678b96b0e8a87505a901a6ea8b452a025f633bb8c5e1b30ca08d9e1bbda8af
+oid sha256:08252754fb4ac87620f3f30d0e4ff704383823ccc54b4b6c040ea20f8e44c40f
 size 167807296

all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -0.8573879599571228,
+    "eval_logits/rejected": -0.7623493671417236,
+    "eval_logps/chosen": -399.9150085449219,
+    "eval_logps/rejected": -371.0875549316406,
+    "eval_loss": 0.05358276888728142,
+    "eval_rewards/accuracies": 0.46950000524520874,
+    "eval_rewards/chosen": -0.003609743434935808,
+    "eval_rewards/margins": 0.0002423849218757823,
+    "eval_rewards/rejected": -0.0038521282840520144,
+    "eval_runtime": 545.4002,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 3.667,
+    "eval_steps_per_second": 0.917,
+    "train_loss": 0.14686254965953338,
+    "train_runtime": 4173.6713,
+    "train_samples": 61135,
+    "train_samples_per_second": 1.198,
+    "train_steps_per_second": 0.075
+}

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 1.0,
+    "eval_logits/chosen": -0.8573879599571228,
+    "eval_logits/rejected": -0.7623493671417236,
+    "eval_logps/chosen": -399.9150085449219,
+    "eval_logps/rejected": -371.0875549316406,
+    "eval_loss": 0.05358276888728142,
+    "eval_rewards/accuracies": 0.46950000524520874,
+    "eval_rewards/chosen": -0.003609743434935808,
+    "eval_rewards/margins": 0.0002423849218757823,
+    "eval_rewards/rejected": -0.0038521282840520144,
+    "eval_runtime": 545.4002,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 3.667,
+    "eval_steps_per_second": 0.917
+}

runs/Apr26_16-30-51_gpu4-119-5/events.out.tfevents.1714113121.gpu4-119-5.1445392.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8c1184b1450274bd2b19f90ace0a8703cc46098d1ab1273cc47281abac0bbca8
-size 26568
+oid sha256:d29757f1200092ebce108e545418e290b35b7348eb8b70477f5a841fbe41ee55
+size 27556

runs/Apr26_16-30-51_gpu4-119-5/events.out.tfevents.1714117840.gpu4-119-5.1445392.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6a3728599c3689f182c7fc04d0f0c40f0f32138b7db2c8ce73b0c0b6be399a6
+size 828

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 1.0,
+    "train_loss": 0.14686254965953338,
+    "train_runtime": 4173.6713,
+    "train_samples": 61135,
+    "train_samples_per_second": 1.198,
+    "train_steps_per_second": 0.075
+}

trainer_state.json

@@ -0,0 +1,526 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9984,
+  "eval_steps": 100,
+  "global_step": 312,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-07,
+      "logits/chosen": -0.6847273707389832,
+      "logits/rejected": -0.5662981271743774,
+      "logps/chosen": -106.9537124633789,
+      "logps/rejected": -85.6178970336914,
+      "loss": 0.1482,
+      "rewards/accuracies": 0.0,
+      "rewards/chosen": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/rejected": 0.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5625e-06,
+      "logits/chosen": -0.7783863544464111,
+      "logits/rejected": -0.7930062413215637,
+      "logps/chosen": -122.19684600830078,
+      "logps/rejected": -112.47706604003906,
+      "loss": 0.1566,
+      "rewards/accuracies": 0.1388888955116272,
+      "rewards/chosen": -0.00023624440655112267,
+      "rewards/margins": -0.00026661725132726133,
+      "rewards/rejected": 3.03728120343294e-05,
+      "step": 10
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.125e-06,
+      "logits/chosen": -0.7614932656288147,
+      "logits/rejected": -0.829719066619873,
+      "logps/chosen": -150.7623748779297,
+      "logps/rejected": -139.49307250976562,
+      "loss": 0.1419,
+      "rewards/accuracies": 0.3375000059604645,
+      "rewards/chosen": 0.0002180041337851435,
+      "rewards/margins": 0.00028577950433827937,
+      "rewards/rejected": -6.77753341733478e-05,
+      "step": 20
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6875000000000004e-06,
+      "logits/chosen": -0.6684157252311707,
+      "logits/rejected": -0.7231897115707397,
+      "logps/chosen": -147.83189392089844,
+      "logps/rejected": -135.98458862304688,
+      "loss": 0.1593,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": -6.6927052102983e-05,
+      "rewards/margins": -4.204309516353533e-05,
+      "rewards/rejected": -2.4883949663490057e-05,
+      "step": 30
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.989935734988098e-06,
+      "logits/chosen": -0.7841319441795349,
+      "logits/rejected": -0.8013907670974731,
+      "logps/chosen": -134.54759216308594,
+      "logps/rejected": -124.2796401977539,
+      "loss": 0.1304,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": -2.134514033969026e-05,
+      "rewards/margins": 3.064397242269479e-05,
+      "rewards/rejected": -5.1989121857332066e-05,
+      "step": 40
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.949188496058089e-06,
+      "logits/chosen": -0.7895469069480896,
+      "logits/rejected": -0.861598789691925,
+      "logps/chosen": -126.7054672241211,
+      "logps/rejected": -113.07292175292969,
+      "loss": 0.1471,
+      "rewards/accuracies": 0.16875000298023224,
+      "rewards/chosen": 0.00019622920081019402,
+      "rewards/margins": -0.0001584202254889533,
+      "rewards/rejected": 0.0003546494117472321,
+      "step": 50
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8776412907378845e-06,
+      "logits/chosen": -0.7303129434585571,
+      "logits/rejected": -0.6805683374404907,
+      "logps/chosen": -136.78421020507812,
+      "logps/rejected": -127.17015075683594,
+      "loss": 0.1519,
+      "rewards/accuracies": 0.2874999940395355,
+      "rewards/chosen": 0.0006400028942152858,
+      "rewards/margins": 3.7805184547323734e-05,
+      "rewards/rejected": 0.0006021977169439197,
+      "step": 60
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7761938666470405e-06,
+      "logits/chosen": -0.8373444676399231,
+      "logits/rejected": -0.737995982170105,
+      "logps/chosen": -131.6265106201172,
+      "logps/rejected": -115.4680404663086,
+      "loss": 0.1346,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": 0.00044487425475381315,
+      "rewards/margins": 5.753149889642373e-05,
+      "rewards/rejected": 0.00038734282134100795,
+      "step": 70
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.646121984004666e-06,
+      "logits/chosen": -0.7500714063644409,
+      "logits/rejected": -0.8881791830062866,
+      "logps/chosen": -129.89303588867188,
+      "logps/rejected": -116.5955810546875,
+      "loss": 0.1493,
+      "rewards/accuracies": 0.26875001192092896,
+      "rewards/chosen": 0.0007193287019617856,
+      "rewards/margins": 0.00013169689918868244,
+      "rewards/rejected": 0.0005876318318769336,
+      "step": 80
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4890613722044526e-06,
+      "logits/chosen": -0.7528460025787354,
+      "logits/rejected": -0.7148691415786743,
+      "logps/chosen": -137.8367462158203,
+      "logps/rejected": -124.02665710449219,
+      "loss": 0.1563,
+      "rewards/accuracies": 0.30000001192092896,
+      "rewards/chosen": 0.0008098526741378009,
+      "rewards/margins": 1.3577568097389303e-05,
+      "rewards/rejected": 0.0007962752133607864,
+      "step": 90
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3069871595684795e-06,
+      "logits/chosen": -0.8201054334640503,
+      "logits/rejected": -0.7307484149932861,
+      "logps/chosen": -129.03594970703125,
+      "logps/rejected": -126.3255844116211,
+      "loss": 0.1203,
+      "rewards/accuracies": 0.33125001192092896,
+      "rewards/chosen": 0.0007475044694729149,
+      "rewards/margins": 0.0003377199172973633,
+      "rewards/rejected": 0.00040978455217555165,
+      "step": 100
+    },
+    {
+      "epoch": 0.32,
+      "eval_logits/chosen": -0.8167486786842346,
+      "eval_logits/rejected": -0.7166811227798462,
+      "eval_logps/chosen": -398.67974853515625,
+      "eval_logps/rejected": -369.66937255859375,
+      "eval_loss": 0.05371898412704468,
+      "eval_rewards/accuracies": 0.4555000066757202,
+      "eval_rewards/chosen": -0.002374465111643076,
+      "eval_rewards/margins": 5.947439422016032e-05,
+      "eval_rewards/rejected": -0.002433939604088664,
+      "eval_runtime": 545.4276,
+      "eval_samples_per_second": 3.667,
+      "eval_steps_per_second": 0.917,
+      "step": 100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.102189034962561e-06,
+      "logits/chosen": -0.7392131090164185,
+      "logits/rejected": -0.8276106119155884,
+      "logps/chosen": -130.80039978027344,
+      "logps/rejected": -132.36734008789062,
+      "loss": 0.1487,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": 0.0011107753962278366,
+      "rewards/margins": 7.14737907401286e-05,
+      "rewards/rejected": 0.001039301510900259,
+      "step": 110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8772424536302565e-06,
+      "logits/chosen": -0.8678170442581177,
+      "logits/rejected": -0.923176109790802,
+      "logps/chosen": -129.24879455566406,
+      "logps/rejected": -120.42958068847656,
+      "loss": 0.1449,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": 0.0010587454307824373,
+      "rewards/margins": 0.00041401953785680234,
+      "rewards/rejected": 0.0006447257474064827,
+      "step": 120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.634976249348867e-06,
+      "logits/chosen": -0.88499915599823,
+      "logits/rejected": -0.8518987894058228,
+      "logps/chosen": -139.58132934570312,
+      "logps/rejected": -131.0221710205078,
+      "loss": 0.1364,
+      "rewards/accuracies": 0.2562499940395355,
+      "rewards/chosen": 0.001031761639751494,
+      "rewards/margins": 0.00010068035044241697,
+      "rewards/rejected": 0.0009310811874456704,
+      "step": 130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.3784370602033572e-06,
+      "logits/chosen": -0.8087924718856812,
+      "logits/rejected": -0.7336855530738831,
+      "logps/chosen": -135.13723754882812,
+      "logps/rejected": -127.080078125,
+      "loss": 0.1654,
+      "rewards/accuracies": 0.26249998807907104,
+      "rewards/chosen": 0.0010691110510379076,
+      "rewards/margins": -2.2428755983128212e-05,
+      "rewards/rejected": 0.0010915396269410849,
+      "step": 140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1108510153447352e-06,
+      "logits/chosen": -0.7911968231201172,
+      "logits/rejected": -0.8422868847846985,
+      "logps/chosen": -129.05540466308594,
+      "logps/rejected": -119.83857727050781,
+      "loss": 0.1523,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": 0.001330427243374288,
+      "rewards/margins": 0.0002512388164177537,
+      "rewards/rejected": 0.0010791884269565344,
+      "step": 150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.835583164544139e-06,
+      "logits/chosen": -0.8172557950019836,
+      "logits/rejected": -0.866910457611084,
+      "logps/chosen": -106.51783752441406,
+      "logps/rejected": -98.27071380615234,
+      "loss": 0.1406,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": 0.0011063208803534508,
+      "rewards/margins": 0.00032531103352084756,
+      "rewards/rejected": 0.0007810097886249423,
+      "step": 160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.556095160739513e-06,
+      "logits/chosen": -0.7146141529083252,
+      "logits/rejected": -0.7281913757324219,
+      "logps/chosen": -141.63967895507812,
+      "logps/rejected": -132.29884338378906,
+      "loss": 0.1425,
+      "rewards/accuracies": 0.3375000059604645,
+      "rewards/chosen": 0.0013892787974327803,
+      "rewards/margins": 7.653595821466297e-05,
+      "rewards/rejected": 0.001312742941081524,
+      "step": 170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2759017277414165e-06,
+      "logits/chosen": -0.7744725942611694,
+      "logits/rejected": -0.7258681058883667,
+      "logps/chosen": -145.58126831054688,
+      "logps/rejected": -130.30874633789062,
+      "loss": 0.1365,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": 0.0011575846001505852,
+      "rewards/margins": -4.4921856897417456e-05,
+      "rewards/rejected": 0.0012025063624605536,
+      "step": 180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9985264605418185e-06,
+      "logits/chosen": -0.8330332636833191,
+      "logits/rejected": -0.7537392377853394,
+      "logps/chosen": -119.37479400634766,
+      "logps/rejected": -117.6336898803711,
+      "loss": 0.1458,
+      "rewards/accuracies": 0.23749999701976776,
+      "rewards/chosen": 0.001272810623049736,
+      "rewards/margins": 1.5811452612979338e-05,
+      "rewards/rejected": 0.0012569990940392017,
+      "step": 190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7274575140626318e-06,
+      "logits/chosen": -0.904527485370636,
+      "logits/rejected": -0.9056881666183472,
+      "logps/chosen": -104.618896484375,
+      "logps/rejected": -98.6053237915039,
+      "loss": 0.1671,
+      "rewards/accuracies": 0.2750000059604645,
+      "rewards/chosen": 0.0011854673502966762,
+      "rewards/margins": 0.00045539819984696805,
+      "rewards/rejected": 0.0007300691795535386,
+      "step": 200
+    },
+    {
+      "epoch": 0.64,
+      "eval_logits/chosen": -0.8673906922340393,
+      "eval_logits/rejected": -0.774476945400238,
+      "eval_logps/chosen": -399.858642578125,
+      "eval_logps/rejected": -370.92401123046875,
+      "eval_loss": 0.053653884679079056,
+      "eval_rewards/accuracies": 0.46700000762939453,
+      "eval_rewards/chosen": -0.0035534072667360306,
+      "eval_rewards/margins": 0.00013517153274733573,
+      "eval_rewards/rejected": -0.003688578959554434,
+      "eval_runtime": 545.4808,
+      "eval_samples_per_second": 3.666,
+      "eval_steps_per_second": 0.917,
+      "step": 200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.466103737583699e-06,
+      "logits/chosen": -0.8533005714416504,
+      "logits/rejected": -0.9184266328811646,
+      "logps/chosen": -138.516357421875,
+      "logps/rejected": -126.01377868652344,
+      "loss": 0.1468,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": 0.0016375569393858314,
+      "rewards/margins": 0.00033169047674164176,
+      "rewards/rejected": 0.0013058666372671723,
+      "step": 210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.217751806485235e-06,
+      "logits/chosen": -0.870014488697052,
+      "logits/rejected": -0.7640275955200195,
+      "logps/chosen": -125.85664367675781,
+      "logps/rejected": -111.66432189941406,
+      "loss": 0.1375,
+      "rewards/accuracies": 0.24375000596046448,
+      "rewards/chosen": 0.0014119090046733618,
+      "rewards/margins": 0.00024046527687460184,
+      "rewards/rejected": 0.0011714436113834381,
+      "step": 220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.855248903979505e-07,
+      "logits/chosen": -0.8237797617912292,
+      "logits/rejected": -0.744855523109436,
+      "logps/chosen": -141.6581268310547,
+      "logps/rejected": -129.85610961914062,
+      "loss": 0.1453,
+      "rewards/accuracies": 0.3687500059604645,
+      "rewards/chosen": 0.0018702894449234009,
+      "rewards/margins": 0.0005412907339632511,
+      "rewards/rejected": 0.001328998594544828,
+      "step": 230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.723433775328385e-07,
+      "logits/chosen": -0.9196878671646118,
+      "logits/rejected": -0.9501888155937195,
+      "logps/chosen": -118.2296142578125,
+      "logps/rejected": -109.28836822509766,
+      "loss": 0.1567,
+      "rewards/accuracies": 0.26249998807907104,
+      "rewards/chosen": 0.0011578628327697515,
+      "rewards/margins": 0.00018040690338239074,
+      "rewards/rejected": 0.0009774559875950217,
+      "step": 240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.808881491049723e-07,
+      "logits/chosen": -0.8927375674247742,
+      "logits/rejected": -0.7841494679450989,
+      "logps/chosen": -148.04861450195312,
+      "logps/rejected": -130.5260772705078,
+      "loss": 0.1682,
+      "rewards/accuracies": 0.33125001192092896,
+      "rewards/chosen": 0.0015076154377311468,
+      "rewards/margins": 0.0004245832678861916,
+      "rewards/rejected": 0.001083032344467938,
+      "step": 250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1356686569674344e-07,
+      "logits/chosen": -0.9602775573730469,
+      "logits/rejected": -0.7984300255775452,
+      "logps/chosen": -114.3084716796875,
+      "logps/rejected": -118.40677642822266,
+      "loss": 0.144,
+      "rewards/accuracies": 0.20624999701976776,
+      "rewards/chosen": 0.0011838999344035983,
+      "rewards/margins": -8.5394291090779e-05,
+      "rewards/rejected": 0.0012692942982539535,
+      "step": 260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7248368952908055e-07,
+      "logits/chosen": -0.7740083932876587,
+      "logits/rejected": -0.835452675819397,
+      "logps/chosen": -143.74681091308594,
+      "logps/rejected": -138.52777099609375,
+      "loss": 0.1722,
+      "rewards/accuracies": 0.32499998807907104,
+      "rewards/chosen": 0.0016226943116635084,
+      "rewards/margins": 0.00025336851831525564,
+      "rewards/rejected": 0.0013693260261788964,
+      "step": 270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.59412823400657e-07,
+      "logits/chosen": -0.8314012289047241,
+      "logits/rejected": -0.7272503972053528,
+      "logps/chosen": -143.94064331054688,
+      "logps/rejected": -130.27163696289062,
+      "loss": 0.1344,
+      "rewards/accuracies": 0.3125,
+      "rewards/chosen": 0.0014273414853960276,
+      "rewards/margins": 0.00031734618823975325,
+      "rewards/rejected": 0.001109995529986918,
+      "step": 280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.577619905828281e-08,
+      "logits/chosen": -0.8207144737243652,
+      "logits/rejected": -0.8345381617546082,
+      "logps/chosen": -150.83251953125,
+      "logps/rejected": -128.1667938232422,
+      "loss": 0.1345,
+      "rewards/accuracies": 0.34375,
+      "rewards/chosen": 0.001841158838942647,
+      "rewards/margins": 0.00017683973419480026,
+      "rewards/rejected": 0.0016643188428133726,
+      "step": 290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.262559558016325e-08,
+      "logits/chosen": -0.8629458546638489,
+      "logits/rejected": -0.7918001413345337,
+      "logps/chosen": -144.69100952148438,
+      "logps/rejected": -134.775634765625,
+      "loss": 0.1393,
+      "rewards/accuracies": 0.3125,
+      "rewards/chosen": 0.0014558725524693727,
+      "rewards/margins": 0.0003581587807275355,
+      "rewards/rejected": 0.0010977135971188545,
+      "step": 300
+    },
+    {
+      "epoch": 0.96,
+      "eval_logits/chosen": -0.8771722316741943,
+      "eval_logits/rejected": -0.781971275806427,
+      "eval_logps/chosen": -400.07305908203125,
+      "eval_logps/rejected": -371.2791442871094,
+      "eval_loss": 0.053590498864650726,
+      "eval_rewards/accuracies": 0.4625000059604645,
+      "eval_rewards/chosen": -0.0037678189110010862,
+      "eval_rewards/margins": 0.0002759249764494598,
+      "eval_rewards/rejected": -0.004043743945658207,
+      "eval_runtime": 545.2618,
+      "eval_samples_per_second": 3.668,
+      "eval_steps_per_second": 0.917,
+      "step": 300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.294126437336734e-10,
+      "logits/chosen": -0.8216972351074219,
+      "logits/rejected": -0.8807767033576965,
+      "logps/chosen": -114.38739013671875,
+      "logps/rejected": -111.93601989746094,
+      "loss": 0.146,
+      "rewards/accuracies": 0.28125,
+      "rewards/chosen": 0.0016306948382407427,
+      "rewards/margins": 0.0005543740699067712,
+      "rewards/rejected": 0.0010763208847492933,
+      "step": 310
+    },
+    {
+      "epoch": 1.0,
+      "step": 312,
+      "total_flos": 0.0,
+      "train_loss": 0.14686254965953338,
+      "train_runtime": 4173.6713,
+      "train_samples_per_second": 1.198,
+      "train_steps_per_second": 0.075
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 312,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}