Model save
Browse files- README.md +61 -0
- all_results.json +9 -0
- generation_config.json +8 -0
- train_results.json +9 -0
- trainer_state.json +447 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: gemma
|
| 4 |
+
base_model: google/gemma-2-9b-it
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- simpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: simpo-train-smallest-30-abs-diff
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# simpo-train-smallest-30-abs-diff
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) on an unknown dataset.
|
| 20 |
+
|
| 21 |
+
## Model description
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Intended uses & limitations
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training and evaluation data
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training procedure
|
| 34 |
+
|
| 35 |
+
### Training hyperparameters
|
| 36 |
+
|
| 37 |
+
The following hyperparameters were used during training:
|
| 38 |
+
- learning_rate: 8e-07
|
| 39 |
+
- train_batch_size: 2
|
| 40 |
+
- eval_batch_size: 4
|
| 41 |
+
- seed: 42
|
| 42 |
+
- distributed_type: multi-GPU
|
| 43 |
+
- num_devices: 4
|
| 44 |
+
- gradient_accumulation_steps: 16
|
| 45 |
+
- total_train_batch_size: 128
|
| 46 |
+
- total_eval_batch_size: 16
|
| 47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
+
- lr_scheduler_type: cosine
|
| 49 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 50 |
+
- num_epochs: 1
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- Transformers 4.44.2
|
| 59 |
+
- Pytorch 2.7.0+cu128
|
| 60 |
+
- Datasets 2.18.0
|
| 61 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.995910949568378,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 4.581099252631194,
|
| 5 |
+
"train_runtime": 3146.1495,
|
| 6 |
+
"train_samples": 17605,
|
| 7 |
+
"train_samples_per_second": 5.596,
|
| 8 |
+
"train_steps_per_second": 0.044
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 2,
|
| 4 |
+
"cache_implementation": "hybrid",
|
| 5 |
+
"eos_token_id": 1,
|
| 6 |
+
"pad_token_id": 0,
|
| 7 |
+
"transformers_version": "4.44.2"
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.995910949568378,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 4.581099252631194,
|
| 5 |
+
"train_runtime": 3146.1495,
|
| 6 |
+
"train_samples": 17605,
|
| 7 |
+
"train_samples_per_second": 5.596,
|
| 8 |
+
"train_steps_per_second": 0.044
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,447 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.995910949568378,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 137,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03634711494775102,
|
| 13 |
+
"grad_norm": 35.987906218257066,
|
| 14 |
+
"learning_rate": 2.857142857142857e-07,
|
| 15 |
+
"logits/chosen": -11.111563682556152,
|
| 16 |
+
"logits/rejected": -11.135547637939453,
|
| 17 |
+
"logps/chosen": -0.662230372428894,
|
| 18 |
+
"logps/rejected": -0.6628919839859009,
|
| 19 |
+
"loss": 5.0952,
|
| 20 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 21 |
+
"rewards/chosen": -6.6223039627075195,
|
| 22 |
+
"rewards/margins": 0.006614901125431061,
|
| 23 |
+
"rewards/rejected": -6.6289191246032715,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.07269422989550205,
|
| 28 |
+
"grad_norm": 28.899244168425763,
|
| 29 |
+
"learning_rate": 5.714285714285714e-07,
|
| 30 |
+
"logits/chosen": -10.829585075378418,
|
| 31 |
+
"logits/rejected": -10.654484748840332,
|
| 32 |
+
"logps/chosen": -0.636147677898407,
|
| 33 |
+
"logps/rejected": -0.6354170441627502,
|
| 34 |
+
"loss": 5.0407,
|
| 35 |
+
"rewards/accuracies": 0.550000011920929,
|
| 36 |
+
"rewards/chosen": -6.361477375030518,
|
| 37 |
+
"rewards/margins": -0.007306712679564953,
|
| 38 |
+
"rewards/rejected": -6.354170799255371,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.10904134484325306,
|
| 43 |
+
"grad_norm": 34.054577665859114,
|
| 44 |
+
"learning_rate": 7.998695344323425e-07,
|
| 45 |
+
"logits/chosen": -10.697937965393066,
|
| 46 |
+
"logits/rejected": -10.68556022644043,
|
| 47 |
+
"logps/chosen": -0.756206214427948,
|
| 48 |
+
"logps/rejected": -0.7363730669021606,
|
| 49 |
+
"loss": 5.0934,
|
| 50 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 51 |
+
"rewards/chosen": -7.562062740325928,
|
| 52 |
+
"rewards/margins": -0.1983323097229004,
|
| 53 |
+
"rewards/rejected": -7.363729953765869,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.1453884597910041,
|
| 58 |
+
"grad_norm": 31.063536076780956,
|
| 59 |
+
"learning_rate": 7.953121695121394e-07,
|
| 60 |
+
"logits/chosen": -10.083427429199219,
|
| 61 |
+
"logits/rejected": -9.985143661499023,
|
| 62 |
+
"logps/chosen": -0.6148006319999695,
|
| 63 |
+
"logps/rejected": -0.6226206421852112,
|
| 64 |
+
"loss": 5.024,
|
| 65 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 66 |
+
"rewards/chosen": -6.148005485534668,
|
| 67 |
+
"rewards/margins": 0.07820063084363937,
|
| 68 |
+
"rewards/rejected": -6.226205825805664,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.18173557473875512,
|
| 73 |
+
"grad_norm": 37.55152432622566,
|
| 74 |
+
"learning_rate": 7.843163833184991e-07,
|
| 75 |
+
"logits/chosen": -9.981136322021484,
|
| 76 |
+
"logits/rejected": -9.983463287353516,
|
| 77 |
+
"logps/chosen": -0.5722960233688354,
|
| 78 |
+
"logps/rejected": -0.5717057585716248,
|
| 79 |
+
"loss": 4.961,
|
| 80 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 81 |
+
"rewards/chosen": -5.722960472106934,
|
| 82 |
+
"rewards/margins": -0.005903184413909912,
|
| 83 |
+
"rewards/rejected": -5.717057228088379,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.21808268968650613,
|
| 88 |
+
"grad_norm": 29.53598266606176,
|
| 89 |
+
"learning_rate": 7.670612634414511e-07,
|
| 90 |
+
"logits/chosen": -9.952229499816895,
|
| 91 |
+
"logits/rejected": -9.804238319396973,
|
| 92 |
+
"logps/chosen": -0.5058998465538025,
|
| 93 |
+
"logps/rejected": -0.5249067544937134,
|
| 94 |
+
"loss": 4.8596,
|
| 95 |
+
"rewards/accuracies": 0.59375,
|
| 96 |
+
"rewards/chosen": -5.058998107910156,
|
| 97 |
+
"rewards/margins": 0.19006863236427307,
|
| 98 |
+
"rewards/rejected": -5.249067306518555,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.25442980463425713,
|
| 103 |
+
"grad_norm": 30.836147418383828,
|
| 104 |
+
"learning_rate": 7.438278427948805e-07,
|
| 105 |
+
"logits/chosen": -9.969837188720703,
|
| 106 |
+
"logits/rejected": -9.884472846984863,
|
| 107 |
+
"logps/chosen": -0.535486102104187,
|
| 108 |
+
"logps/rejected": -0.5739948153495789,
|
| 109 |
+
"loss": 4.7794,
|
| 110 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 111 |
+
"rewards/chosen": -5.354860782623291,
|
| 112 |
+
"rewards/margins": 0.3850872814655304,
|
| 113 |
+
"rewards/rejected": -5.739948272705078,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2907769195820082,
|
| 118 |
+
"grad_norm": 35.784369740177794,
|
| 119 |
+
"learning_rate": 7.149945224533862e-07,
|
| 120 |
+
"logits/chosen": -9.953130722045898,
|
| 121 |
+
"logits/rejected": -9.866512298583984,
|
| 122 |
+
"logps/chosen": -0.6358538866043091,
|
| 123 |
+
"logps/rejected": -0.65082848072052,
|
| 124 |
+
"loss": 4.8879,
|
| 125 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 126 |
+
"rewards/chosen": -6.358539581298828,
|
| 127 |
+
"rewards/margins": 0.14974501729011536,
|
| 128 |
+
"rewards/rejected": -6.508284091949463,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3271240345297592,
|
| 133 |
+
"grad_norm": 30.738414397371837,
|
| 134 |
+
"learning_rate": 6.810309086608129e-07,
|
| 135 |
+
"logits/chosen": -10.294210433959961,
|
| 136 |
+
"logits/rejected": -10.260662078857422,
|
| 137 |
+
"logps/chosen": -0.6186165809631348,
|
| 138 |
+
"logps/rejected": -0.642761766910553,
|
| 139 |
+
"loss": 4.817,
|
| 140 |
+
"rewards/accuracies": 0.581250011920929,
|
| 141 |
+
"rewards/chosen": -6.186165809631348,
|
| 142 |
+
"rewards/margins": 0.24145162105560303,
|
| 143 |
+
"rewards/rejected": -6.427617073059082,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.36347114947751025,
|
| 148 |
+
"grad_norm": 46.12276326749997,
|
| 149 |
+
"learning_rate": 6.424901643866552e-07,
|
| 150 |
+
"logits/chosen": -10.330435752868652,
|
| 151 |
+
"logits/rejected": -10.143304824829102,
|
| 152 |
+
"logps/chosen": -0.6831095814704895,
|
| 153 |
+
"logps/rejected": -0.7464486956596375,
|
| 154 |
+
"loss": 4.6885,
|
| 155 |
+
"rewards/accuracies": 0.6875,
|
| 156 |
+
"rewards/chosen": -6.8310956954956055,
|
| 157 |
+
"rewards/margins": 0.6333915591239929,
|
| 158 |
+
"rewards/rejected": -7.464486598968506,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.39981826442526125,
|
| 163 |
+
"grad_norm": 43.40349458380626,
|
| 164 |
+
"learning_rate": 6e-07,
|
| 165 |
+
"logits/chosen": -11.197371482849121,
|
| 166 |
+
"logits/rejected": -11.12946605682373,
|
| 167 |
+
"logps/chosen": -0.5718105435371399,
|
| 168 |
+
"logps/rejected": -0.6250694394111633,
|
| 169 |
+
"loss": 4.6953,
|
| 170 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 171 |
+
"rewards/chosen": -5.718105792999268,
|
| 172 |
+
"rewards/margins": 0.5325883626937866,
|
| 173 |
+
"rewards/rejected": -6.250694274902344,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.43616537937301225,
|
| 178 |
+
"grad_norm": 62.775385851545494,
|
| 179 |
+
"learning_rate": 5.542524497952543e-07,
|
| 180 |
+
"logits/chosen": -10.749040603637695,
|
| 181 |
+
"logits/rejected": -10.889803886413574,
|
| 182 |
+
"logps/chosen": -0.6364445686340332,
|
| 183 |
+
"logps/rejected": -0.7200480699539185,
|
| 184 |
+
"loss": 4.6271,
|
| 185 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 186 |
+
"rewards/chosen": -6.364445686340332,
|
| 187 |
+
"rewards/margins": 0.8360347747802734,
|
| 188 |
+
"rewards/rejected": -7.200479984283447,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.4725124943207633,
|
| 193 |
+
"grad_norm": 45.98517277447237,
|
| 194 |
+
"learning_rate": 5.059926008786647e-07,
|
| 195 |
+
"logits/chosen": -11.873740196228027,
|
| 196 |
+
"logits/rejected": -11.824403762817383,
|
| 197 |
+
"logps/chosen": -0.7099884748458862,
|
| 198 |
+
"logps/rejected": -0.7525407075881958,
|
| 199 |
+
"loss": 4.638,
|
| 200 |
+
"rewards/accuracies": 0.59375,
|
| 201 |
+
"rewards/chosen": -7.0998854637146,
|
| 202 |
+
"rewards/margins": 0.4255215525627136,
|
| 203 |
+
"rewards/rejected": -7.525406837463379,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.5088596092685143,
|
| 208 |
+
"grad_norm": 47.30579548391997,
|
| 209 |
+
"learning_rate": 4.5600645798745166e-07,
|
| 210 |
+
"logits/chosen": -12.138381004333496,
|
| 211 |
+
"logits/rejected": -11.983818054199219,
|
| 212 |
+
"logps/chosen": -0.6762068867683411,
|
| 213 |
+
"logps/rejected": -0.7381676435470581,
|
| 214 |
+
"loss": 4.703,
|
| 215 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 216 |
+
"rewards/chosen": -6.762068748474121,
|
| 217 |
+
"rewards/margins": 0.619607150554657,
|
| 218 |
+
"rewards/rejected": -7.38167667388916,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.5452067242162654,
|
| 223 |
+
"grad_norm": 69.5932279945552,
|
| 224 |
+
"learning_rate": 4.051081418863895e-07,
|
| 225 |
+
"logits/chosen": -12.561568260192871,
|
| 226 |
+
"logits/rejected": -12.478489875793457,
|
| 227 |
+
"logps/chosen": -0.6981841325759888,
|
| 228 |
+
"logps/rejected": -0.7842732667922974,
|
| 229 |
+
"loss": 4.5323,
|
| 230 |
+
"rewards/accuracies": 0.643750011920929,
|
| 231 |
+
"rewards/chosen": -6.98184061050415,
|
| 232 |
+
"rewards/margins": 0.8608924150466919,
|
| 233 |
+
"rewards/rejected": -7.842732906341553,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5815538391640164,
|
| 238 |
+
"grad_norm": 127.53517276446773,
|
| 239 |
+
"learning_rate": 3.541266298406398e-07,
|
| 240 |
+
"logits/chosen": -12.076669692993164,
|
| 241 |
+
"logits/rejected": -12.142119407653809,
|
| 242 |
+
"logps/chosen": -0.7188806533813477,
|
| 243 |
+
"logps/rejected": -0.7781528234481812,
|
| 244 |
+
"loss": 4.6379,
|
| 245 |
+
"rewards/accuracies": 0.581250011920929,
|
| 246 |
+
"rewards/chosen": -7.188807010650635,
|
| 247 |
+
"rewards/margins": 0.5927222371101379,
|
| 248 |
+
"rewards/rejected": -7.781529426574707,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.6179009541117674,
|
| 253 |
+
"grad_norm": 61.22401801870787,
|
| 254 |
+
"learning_rate": 3.0389225412181565e-07,
|
| 255 |
+
"logits/chosen": -13.094499588012695,
|
| 256 |
+
"logits/rejected": -13.159561157226562,
|
| 257 |
+
"logps/chosen": -0.7800406813621521,
|
| 258 |
+
"logps/rejected": -0.8725376129150391,
|
| 259 |
+
"loss": 4.4548,
|
| 260 |
+
"rewards/accuracies": 0.637499988079071,
|
| 261 |
+
"rewards/chosen": -7.800407409667969,
|
| 262 |
+
"rewards/margins": 0.9249688386917114,
|
| 263 |
+
"rewards/rejected": -8.725375175476074,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.6542480690595184,
|
| 268 |
+
"grad_norm": 66.37522679776315,
|
| 269 |
+
"learning_rate": 2.5522317844515273e-07,
|
| 270 |
+
"logits/chosen": -13.562166213989258,
|
| 271 |
+
"logits/rejected": -13.433174133300781,
|
| 272 |
+
"logps/chosen": -0.7740485072135925,
|
| 273 |
+
"logps/rejected": -0.8862950205802917,
|
| 274 |
+
"loss": 4.3623,
|
| 275 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 276 |
+
"rewards/chosen": -7.740484714508057,
|
| 277 |
+
"rewards/margins": 1.1224651336669922,
|
| 278 |
+
"rewards/rejected": -8.862950325012207,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.6905951840072694,
|
| 283 |
+
"grad_norm": 72.9710577397891,
|
| 284 |
+
"learning_rate": 2.0891207259509476e-07,
|
| 285 |
+
"logits/chosen": -14.113569259643555,
|
| 286 |
+
"logits/rejected": -13.75220012664795,
|
| 287 |
+
"logps/chosen": -0.7652403116226196,
|
| 288 |
+
"logps/rejected": -0.8429893255233765,
|
| 289 |
+
"loss": 4.3962,
|
| 290 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 291 |
+
"rewards/chosen": -7.652403354644775,
|
| 292 |
+
"rewards/margins": 0.7774893045425415,
|
| 293 |
+
"rewards/rejected": -8.429892539978027,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7269422989550205,
|
| 298 |
+
"grad_norm": 60.63378631631187,
|
| 299 |
+
"learning_rate": 1.6571320226872206e-07,
|
| 300 |
+
"logits/chosen": -14.494039535522461,
|
| 301 |
+
"logits/rejected": -14.628949165344238,
|
| 302 |
+
"logps/chosen": -0.7868660688400269,
|
| 303 |
+
"logps/rejected": -0.9212242960929871,
|
| 304 |
+
"loss": 4.1386,
|
| 305 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 306 |
+
"rewards/chosen": -7.868659973144531,
|
| 307 |
+
"rewards/margins": 1.3435838222503662,
|
| 308 |
+
"rewards/rejected": -9.212244033813477,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.7632894139027715,
|
| 313 |
+
"grad_norm": 190.4622656460041,
|
| 314 |
+
"learning_rate": 1.2633014440382787e-07,
|
| 315 |
+
"logits/chosen": -15.230148315429688,
|
| 316 |
+
"logits/rejected": -14.929168701171875,
|
| 317 |
+
"logps/chosen": -0.8822343945503235,
|
| 318 |
+
"logps/rejected": -0.9908519983291626,
|
| 319 |
+
"loss": 4.2736,
|
| 320 |
+
"rewards/accuracies": 0.612500011920929,
|
| 321 |
+
"rewards/chosen": -8.822343826293945,
|
| 322 |
+
"rewards/margins": 1.0861766338348389,
|
| 323 |
+
"rewards/rejected": -9.908520698547363,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.7996365288505225,
|
| 328 |
+
"grad_norm": 101.38847574260835,
|
| 329 |
+
"learning_rate": 9.14043280712228e-08,
|
| 330 |
+
"logits/chosen": -15.496833801269531,
|
| 331 |
+
"logits/rejected": -15.636553764343262,
|
| 332 |
+
"logps/chosen": -0.8570655584335327,
|
| 333 |
+
"logps/rejected": -0.9334220886230469,
|
| 334 |
+
"loss": 4.2832,
|
| 335 |
+
"rewards/accuracies": 0.668749988079071,
|
| 336 |
+
"rewards/chosen": -8.570656776428223,
|
| 337 |
+
"rewards/margins": 0.7635642886161804,
|
| 338 |
+
"rewards/rejected": -9.334220886230469,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8359836437982735,
|
| 343 |
+
"grad_norm": 74.21593181905395,
|
| 344 |
+
"learning_rate": 6.150458756494239e-08,
|
| 345 |
+
"logits/chosen": -15.494606018066406,
|
| 346 |
+
"logits/rejected": -15.465006828308105,
|
| 347 |
+
"logps/chosen": -0.9054273366928101,
|
| 348 |
+
"logps/rejected": -1.0216588973999023,
|
| 349 |
+
"loss": 4.1426,
|
| 350 |
+
"rewards/accuracies": 0.668749988079071,
|
| 351 |
+
"rewards/chosen": -9.05427360534668,
|
| 352 |
+
"rewards/margins": 1.1623156070709229,
|
| 353 |
+
"rewards/rejected": -10.216588973999023,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.8723307587460245,
|
| 358 |
+
"grad_norm": 65.9311757106038,
|
| 359 |
+
"learning_rate": 3.711789783843522e-08,
|
| 360 |
+
"logits/chosen": -15.544687271118164,
|
| 361 |
+
"logits/rejected": -15.537959098815918,
|
| 362 |
+
"logps/chosen": -0.8991163969039917,
|
| 363 |
+
"logps/rejected": -0.9694231748580933,
|
| 364 |
+
"loss": 4.2329,
|
| 365 |
+
"rewards/accuracies": 0.637499988079071,
|
| 366 |
+
"rewards/chosen": -8.991164207458496,
|
| 367 |
+
"rewards/margins": 0.7030680179595947,
|
| 368 |
+
"rewards/rejected": -9.694231033325195,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9086778736937755,
|
| 373 |
+
"grad_norm": 130.83461356205703,
|
| 374 |
+
"learning_rate": 1.8641443178027784e-08,
|
| 375 |
+
"logits/chosen": -15.450345039367676,
|
| 376 |
+
"logits/rejected": -15.695470809936523,
|
| 377 |
+
"logps/chosen": -0.8895367383956909,
|
| 378 |
+
"logps/rejected": -1.0279381275177002,
|
| 379 |
+
"loss": 4.1031,
|
| 380 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 381 |
+
"rewards/chosen": -8.895367622375488,
|
| 382 |
+
"rewards/margins": 1.3840134143829346,
|
| 383 |
+
"rewards/rejected": -10.279379844665527,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.9450249886415266,
|
| 388 |
+
"grad_norm": 82.30298232430259,
|
| 389 |
+
"learning_rate": 6.376148290617145e-09,
|
| 390 |
+
"logits/chosen": -16.2006778717041,
|
| 391 |
+
"logits/rejected": -16.138973236083984,
|
| 392 |
+
"logps/chosen": -0.9445899724960327,
|
| 393 |
+
"logps/rejected": -1.0684407949447632,
|
| 394 |
+
"loss": 4.0298,
|
| 395 |
+
"rewards/accuracies": 0.699999988079071,
|
| 396 |
+
"rewards/chosen": -9.445899963378906,
|
| 397 |
+
"rewards/margins": 1.2385092973709106,
|
| 398 |
+
"rewards/rejected": -10.684408187866211,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.9813721035892776,
|
| 403 |
+
"grad_norm": 88.42391053771068,
|
| 404 |
+
"learning_rate": 5.217771643080127e-10,
|
| 405 |
+
"logits/chosen": -16.4208927154541,
|
| 406 |
+
"logits/rejected": -16.113683700561523,
|
| 407 |
+
"logps/chosen": -0.8756990432739258,
|
| 408 |
+
"logps/rejected": -0.9942779541015625,
|
| 409 |
+
"loss": 4.1263,
|
| 410 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 411 |
+
"rewards/chosen": -8.756990432739258,
|
| 412 |
+
"rewards/margins": 1.1857887506484985,
|
| 413 |
+
"rewards/rejected": -9.942779541015625,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.995910949568378,
|
| 418 |
+
"step": 137,
|
| 419 |
+
"total_flos": 0.0,
|
| 420 |
+
"train_loss": 4.581099252631194,
|
| 421 |
+
"train_runtime": 3146.1495,
|
| 422 |
+
"train_samples_per_second": 5.596,
|
| 423 |
+
"train_steps_per_second": 0.044
|
| 424 |
+
}
|
| 425 |
+
],
|
| 426 |
+
"logging_steps": 5,
|
| 427 |
+
"max_steps": 137,
|
| 428 |
+
"num_input_tokens_seen": 0,
|
| 429 |
+
"num_train_epochs": 1,
|
| 430 |
+
"save_steps": 1000000,
|
| 431 |
+
"stateful_callbacks": {
|
| 432 |
+
"TrainerControl": {
|
| 433 |
+
"args": {
|
| 434 |
+
"should_epoch_stop": false,
|
| 435 |
+
"should_evaluate": false,
|
| 436 |
+
"should_log": false,
|
| 437 |
+
"should_save": true,
|
| 438 |
+
"should_training_stop": true
|
| 439 |
+
},
|
| 440 |
+
"attributes": {}
|
| 441 |
+
}
|
| 442 |
+
},
|
| 443 |
+
"total_flos": 0.0,
|
| 444 |
+
"train_batch_size": 2,
|
| 445 |
+
"trial_name": null,
|
| 446 |
+
"trial_params": null
|
| 447 |
+
}
|