Model save
Browse files- README.md +74 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +358 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-align-scan-0.0-0.0-polynomial-2
|
| 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 |
+
# zephyr-7b-align-scan-0.0-0.0-polynomial-2
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6444
|
| 22 |
+
- Rewards/chosen: -0.1722
|
| 23 |
+
- Rewards/rejected: -0.3813
|
| 24 |
+
- Rewards/accuracies: 0.3234
|
| 25 |
+
- Rewards/margins: 0.2091
|
| 26 |
+
- Logps/rejected: -91.3032
|
| 27 |
+
- Logps/chosen: -79.0868
|
| 28 |
+
- Logits/rejected: -2.3982
|
| 29 |
+
- Logits/chosen: -2.4166
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 4.045909641846537e-07
|
| 49 |
+
- train_batch_size: 8
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 2
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 32
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: polynomial
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 2
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.5735 | 1.0417 | 100 | 0.6444 | -0.1722 | -0.3813 | 0.3234 | 0.2091 | -91.3032 | -79.0868 | -2.3982 | -2.4166 |
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
### Framework versions
|
| 70 |
+
|
| 71 |
+
- Transformers 4.44.2
|
| 72 |
+
- Pytorch 2.4.0
|
| 73 |
+
- Datasets 2.21.0
|
| 74 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5773687552039822,
|
| 5 |
+
"train_runtime": 2519.8607,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 4.852,
|
| 8 |
+
"train_steps_per_second": 0.076
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.2"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5773687552039822,
|
| 5 |
+
"train_runtime": 2519.8607,
|
| 6 |
+
"train_samples": 6113,
|
| 7 |
+
"train_samples_per_second": 4.852,
|
| 8 |
+
"train_steps_per_second": 0.076
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,358 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 192,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010416666666666666,
|
| 13 |
+
"grad_norm": 24.66960228608589,
|
| 14 |
+
"learning_rate": 2.0229548209232687e-08,
|
| 15 |
+
"logits/chosen": -2.590585231781006,
|
| 16 |
+
"logits/rejected": -2.5664222240448,
|
| 17 |
+
"logps/chosen": -80.29847717285156,
|
| 18 |
+
"logps/rejected": -53.10200881958008,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.10416666666666667,
|
| 28 |
+
"grad_norm": 22.00841805745166,
|
| 29 |
+
"learning_rate": 2.0229548209232686e-07,
|
| 30 |
+
"logits/chosen": -2.5565450191497803,
|
| 31 |
+
"logits/rejected": -2.5385091304779053,
|
| 32 |
+
"logps/chosen": -87.94749450683594,
|
| 33 |
+
"logps/rejected": -81.03849792480469,
|
| 34 |
+
"loss": 0.6933,
|
| 35 |
+
"rewards/accuracies": 0.1666666716337204,
|
| 36 |
+
"rewards/chosen": -0.0012115496210753918,
|
| 37 |
+
"rewards/margins": -0.001211885130032897,
|
| 38 |
+
"rewards/rejected": 3.3554130141055794e-07,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.20833333333333334,
|
| 43 |
+
"grad_norm": 19.440282490720026,
|
| 44 |
+
"learning_rate": 4.045909641846537e-07,
|
| 45 |
+
"logits/chosen": -2.612945318222046,
|
| 46 |
+
"logits/rejected": -2.5642082691192627,
|
| 47 |
+
"logps/chosen": -103.17449951171875,
|
| 48 |
+
"logps/rejected": -89.7829360961914,
|
| 49 |
+
"loss": 0.6911,
|
| 50 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 51 |
+
"rewards/chosen": 0.006612158380448818,
|
| 52 |
+
"rewards/margins": 0.0028112882282584906,
|
| 53 |
+
"rewards/rejected": 0.0038008708506822586,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.3125,
|
| 58 |
+
"grad_norm": 25.805649105098475,
|
| 59 |
+
"learning_rate": 3.8688218719717384e-07,
|
| 60 |
+
"logits/chosen": -2.508690357208252,
|
| 61 |
+
"logits/rejected": -2.5234837532043457,
|
| 62 |
+
"logps/chosen": -66.37418365478516,
|
| 63 |
+
"logps/rejected": -74.30455017089844,
|
| 64 |
+
"loss": 0.6851,
|
| 65 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 66 |
+
"rewards/chosen": 0.03983448073267937,
|
| 67 |
+
"rewards/margins": 0.016553020104765892,
|
| 68 |
+
"rewards/rejected": 0.023281460627913475,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.4166666666666667,
|
| 73 |
+
"grad_norm": 17.133204904463383,
|
| 74 |
+
"learning_rate": 3.6917341020969396e-07,
|
| 75 |
+
"logits/chosen": -2.555058479309082,
|
| 76 |
+
"logits/rejected": -2.5424702167510986,
|
| 77 |
+
"logps/chosen": -69.98268127441406,
|
| 78 |
+
"logps/rejected": -69.5423583984375,
|
| 79 |
+
"loss": 0.6777,
|
| 80 |
+
"rewards/accuracies": 0.28125,
|
| 81 |
+
"rewards/chosen": 0.08108103275299072,
|
| 82 |
+
"rewards/margins": 0.03720525652170181,
|
| 83 |
+
"rewards/rejected": 0.04387578368186951,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.5208333333333334,
|
| 88 |
+
"grad_norm": 22.625402130027865,
|
| 89 |
+
"learning_rate": 3.5146463322221414e-07,
|
| 90 |
+
"logits/chosen": -2.4543604850769043,
|
| 91 |
+
"logits/rejected": -2.467179775238037,
|
| 92 |
+
"logps/chosen": -49.026615142822266,
|
| 93 |
+
"logps/rejected": -58.0882568359375,
|
| 94 |
+
"loss": 0.6703,
|
| 95 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 96 |
+
"rewards/chosen": 0.07023879140615463,
|
| 97 |
+
"rewards/margins": 0.03948055952787399,
|
| 98 |
+
"rewards/rejected": 0.030758222565054893,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.625,
|
| 103 |
+
"grad_norm": 23.61323531343505,
|
| 104 |
+
"learning_rate": 3.3375585623473426e-07,
|
| 105 |
+
"logits/chosen": -2.530721664428711,
|
| 106 |
+
"logits/rejected": -2.5115602016448975,
|
| 107 |
+
"logps/chosen": -78.50758361816406,
|
| 108 |
+
"logps/rejected": -81.587158203125,
|
| 109 |
+
"loss": 0.6625,
|
| 110 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 111 |
+
"rewards/chosen": 0.02104257419705391,
|
| 112 |
+
"rewards/margins": 0.1053396612405777,
|
| 113 |
+
"rewards/rejected": -0.08429709076881409,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7291666666666666,
|
| 118 |
+
"grad_norm": 21.448302962597314,
|
| 119 |
+
"learning_rate": 3.1604707924725434e-07,
|
| 120 |
+
"logits/chosen": -2.4924867153167725,
|
| 121 |
+
"logits/rejected": -2.4797446727752686,
|
| 122 |
+
"logps/chosen": -92.79881286621094,
|
| 123 |
+
"logps/rejected": -86.05406188964844,
|
| 124 |
+
"loss": 0.6497,
|
| 125 |
+
"rewards/accuracies": 0.35624998807907104,
|
| 126 |
+
"rewards/chosen": 0.13287237286567688,
|
| 127 |
+
"rewards/margins": 0.15550200641155243,
|
| 128 |
+
"rewards/rejected": -0.02262965776026249,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.8333333333333334,
|
| 133 |
+
"grad_norm": 34.07385900953939,
|
| 134 |
+
"learning_rate": 2.983383022597745e-07,
|
| 135 |
+
"logits/chosen": -2.470043182373047,
|
| 136 |
+
"logits/rejected": -2.412937879562378,
|
| 137 |
+
"logps/chosen": -82.65133666992188,
|
| 138 |
+
"logps/rejected": -82.9260482788086,
|
| 139 |
+
"loss": 0.636,
|
| 140 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 141 |
+
"rewards/chosen": 0.11218090355396271,
|
| 142 |
+
"rewards/margins": 0.24496188759803772,
|
| 143 |
+
"rewards/rejected": -0.13278096914291382,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.9375,
|
| 148 |
+
"grad_norm": 29.698805422230546,
|
| 149 |
+
"learning_rate": 2.806295252722946e-07,
|
| 150 |
+
"logits/chosen": -2.402101755142212,
|
| 151 |
+
"logits/rejected": -2.3976664543151855,
|
| 152 |
+
"logps/chosen": -55.17719650268555,
|
| 153 |
+
"logps/rejected": -69.18394470214844,
|
| 154 |
+
"loss": 0.6461,
|
| 155 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 156 |
+
"rewards/chosen": -0.04486609995365143,
|
| 157 |
+
"rewards/margins": 0.1693090796470642,
|
| 158 |
+
"rewards/rejected": -0.21417517960071564,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.0416666666666667,
|
| 163 |
+
"grad_norm": 22.229206377048136,
|
| 164 |
+
"learning_rate": 2.6292074828481476e-07,
|
| 165 |
+
"logits/chosen": -2.388129711151123,
|
| 166 |
+
"logits/rejected": -2.3694980144500732,
|
| 167 |
+
"logps/chosen": -71.2069320678711,
|
| 168 |
+
"logps/rejected": -80.87518310546875,
|
| 169 |
+
"loss": 0.5735,
|
| 170 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 171 |
+
"rewards/chosen": 0.08674298971891403,
|
| 172 |
+
"rewards/margins": 0.4183635711669922,
|
| 173 |
+
"rewards/rejected": -0.33162060379981995,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.0416666666666667,
|
| 178 |
+
"eval_logits/chosen": -2.4165902137756348,
|
| 179 |
+
"eval_logits/rejected": -2.3982186317443848,
|
| 180 |
+
"eval_logps/chosen": -79.08678436279297,
|
| 181 |
+
"eval_logps/rejected": -91.30322265625,
|
| 182 |
+
"eval_loss": 0.6444148421287537,
|
| 183 |
+
"eval_rewards/accuracies": 0.32341268658638,
|
| 184 |
+
"eval_rewards/chosen": -0.1721993237733841,
|
| 185 |
+
"eval_rewards/margins": 0.2090616375207901,
|
| 186 |
+
"eval_rewards/rejected": -0.3812609612941742,
|
| 187 |
+
"eval_runtime": 113.7367,
|
| 188 |
+
"eval_samples_per_second": 17.584,
|
| 189 |
+
"eval_steps_per_second": 0.554,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.1458333333333333,
|
| 194 |
+
"grad_norm": 18.970483409354923,
|
| 195 |
+
"learning_rate": 2.452119712973349e-07,
|
| 196 |
+
"logits/chosen": -2.379981279373169,
|
| 197 |
+
"logits/rejected": -2.420714855194092,
|
| 198 |
+
"logps/chosen": -61.6137809753418,
|
| 199 |
+
"logps/rejected": -94.87492370605469,
|
| 200 |
+
"loss": 0.5017,
|
| 201 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 202 |
+
"rewards/chosen": 0.1406974196434021,
|
| 203 |
+
"rewards/margins": 0.6424610018730164,
|
| 204 |
+
"rewards/rejected": -0.5017635822296143,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.25,
|
| 209 |
+
"grad_norm": 25.179155759555773,
|
| 210 |
+
"learning_rate": 2.27503194309855e-07,
|
| 211 |
+
"logits/chosen": -2.4509730339050293,
|
| 212 |
+
"logits/rejected": -2.412595510482788,
|
| 213 |
+
"logps/chosen": -103.92436218261719,
|
| 214 |
+
"logps/rejected": -112.41646575927734,
|
| 215 |
+
"loss": 0.5125,
|
| 216 |
+
"rewards/accuracies": 0.512499988079071,
|
| 217 |
+
"rewards/chosen": 0.1529318243265152,
|
| 218 |
+
"rewards/margins": 0.8678812980651855,
|
| 219 |
+
"rewards/rejected": -0.7149494886398315,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.3541666666666667,
|
| 224 |
+
"grad_norm": 24.005615665444957,
|
| 225 |
+
"learning_rate": 2.0979441732237514e-07,
|
| 226 |
+
"logits/chosen": -2.365518093109131,
|
| 227 |
+
"logits/rejected": -2.3705790042877197,
|
| 228 |
+
"logps/chosen": -82.37178802490234,
|
| 229 |
+
"logps/rejected": -104.2152328491211,
|
| 230 |
+
"loss": 0.4926,
|
| 231 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 232 |
+
"rewards/chosen": 0.15715977549552917,
|
| 233 |
+
"rewards/margins": 0.7146759033203125,
|
| 234 |
+
"rewards/rejected": -0.5575161576271057,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.4583333333333333,
|
| 239 |
+
"grad_norm": 30.460528822298023,
|
| 240 |
+
"learning_rate": 1.9208564033489532e-07,
|
| 241 |
+
"logits/chosen": -2.3874213695526123,
|
| 242 |
+
"logits/rejected": -2.410029888153076,
|
| 243 |
+
"logps/chosen": -52.55329513549805,
|
| 244 |
+
"logps/rejected": -84.75013732910156,
|
| 245 |
+
"loss": 0.5088,
|
| 246 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 247 |
+
"rewards/chosen": -0.12236322462558746,
|
| 248 |
+
"rewards/margins": 0.6131333708763123,
|
| 249 |
+
"rewards/rejected": -0.7354966402053833,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.5625,
|
| 254 |
+
"grad_norm": 23.593513513479692,
|
| 255 |
+
"learning_rate": 1.7437686334741544e-07,
|
| 256 |
+
"logits/chosen": -2.447330951690674,
|
| 257 |
+
"logits/rejected": -2.4216654300689697,
|
| 258 |
+
"logps/chosen": -73.29743957519531,
|
| 259 |
+
"logps/rejected": -96.33416748046875,
|
| 260 |
+
"loss": 0.4942,
|
| 261 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 262 |
+
"rewards/chosen": -0.04307596758008003,
|
| 263 |
+
"rewards/margins": 0.809691309928894,
|
| 264 |
+
"rewards/rejected": -0.852767288684845,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.6666666666666665,
|
| 269 |
+
"grad_norm": 30.463548320289735,
|
| 270 |
+
"learning_rate": 1.5666808635993556e-07,
|
| 271 |
+
"logits/chosen": -2.4006917476654053,
|
| 272 |
+
"logits/rejected": -2.410930871963501,
|
| 273 |
+
"logps/chosen": -77.80570983886719,
|
| 274 |
+
"logps/rejected": -112.13191223144531,
|
| 275 |
+
"loss": 0.4833,
|
| 276 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 277 |
+
"rewards/chosen": 0.116888128221035,
|
| 278 |
+
"rewards/margins": 1.0234750509262085,
|
| 279 |
+
"rewards/rejected": -0.9065868258476257,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.7708333333333335,
|
| 284 |
+
"grad_norm": 21.813576335226717,
|
| 285 |
+
"learning_rate": 1.389593093724557e-07,
|
| 286 |
+
"logits/chosen": -2.349498748779297,
|
| 287 |
+
"logits/rejected": -2.345773458480835,
|
| 288 |
+
"logps/chosen": -75.15486145019531,
|
| 289 |
+
"logps/rejected": -106.83766174316406,
|
| 290 |
+
"loss": 0.4879,
|
| 291 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 292 |
+
"rewards/chosen": 0.00244431896135211,
|
| 293 |
+
"rewards/margins": 0.8961756825447083,
|
| 294 |
+
"rewards/rejected": -0.893731415271759,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.875,
|
| 299 |
+
"grad_norm": 24.30833424266402,
|
| 300 |
+
"learning_rate": 1.2125053238497584e-07,
|
| 301 |
+
"logits/chosen": -2.3641536235809326,
|
| 302 |
+
"logits/rejected": -2.3408045768737793,
|
| 303 |
+
"logps/chosen": -69.39147186279297,
|
| 304 |
+
"logps/rejected": -93.45851135253906,
|
| 305 |
+
"loss": 0.467,
|
| 306 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 307 |
+
"rewards/chosen": 0.09233461320400238,
|
| 308 |
+
"rewards/margins": 0.8359875679016113,
|
| 309 |
+
"rewards/rejected": -0.7436529397964478,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.9791666666666665,
|
| 314 |
+
"grad_norm": 22.898056429808634,
|
| 315 |
+
"learning_rate": 1.0354175539749597e-07,
|
| 316 |
+
"logits/chosen": -2.3930718898773193,
|
| 317 |
+
"logits/rejected": -2.3902525901794434,
|
| 318 |
+
"logps/chosen": -86.51902770996094,
|
| 319 |
+
"logps/rejected": -118.46821594238281,
|
| 320 |
+
"loss": 0.4742,
|
| 321 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 322 |
+
"rewards/chosen": 0.07279382646083832,
|
| 323 |
+
"rewards/margins": 1.116233229637146,
|
| 324 |
+
"rewards/rejected": -1.0434393882751465,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.0,
|
| 329 |
+
"step": 192,
|
| 330 |
+
"total_flos": 0.0,
|
| 331 |
+
"train_loss": 0.5773687552039822,
|
| 332 |
+
"train_runtime": 2519.8607,
|
| 333 |
+
"train_samples_per_second": 4.852,
|
| 334 |
+
"train_steps_per_second": 0.076
|
| 335 |
+
}
|
| 336 |
+
],
|
| 337 |
+
"logging_steps": 10,
|
| 338 |
+
"max_steps": 192,
|
| 339 |
+
"num_input_tokens_seen": 0,
|
| 340 |
+
"num_train_epochs": 2,
|
| 341 |
+
"save_steps": 100,
|
| 342 |
+
"stateful_callbacks": {
|
| 343 |
+
"TrainerControl": {
|
| 344 |
+
"args": {
|
| 345 |
+
"should_epoch_stop": false,
|
| 346 |
+
"should_evaluate": false,
|
| 347 |
+
"should_log": false,
|
| 348 |
+
"should_save": true,
|
| 349 |
+
"should_training_stop": true
|
| 350 |
+
},
|
| 351 |
+
"attributes": {}
|
| 352 |
+
}
|
| 353 |
+
},
|
| 354 |
+
"total_flos": 0.0,
|
| 355 |
+
"train_batch_size": 8,
|
| 356 |
+
"trial_name": null,
|
| 357 |
+
"trial_params": null
|
| 358 |
+
}
|