Model save
Browse files- README.md +68 -0
- all_results.json +8 -0
- generation_config.json +6 -0
- train_results.json +8 -0
- trainer_state.json +512 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: Qwen/Qwen2.5-7B
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: Qwen-2.5-Base-7B-gen8-math3to5_olympiads_aime-ghpo-cold0-hint50-prompt1-redonum-test
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- ghpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for Qwen-2.5-Base-7B-gen8-math3to5_olympiads_aime-ghpo-cold0-hint50-prompt1-redonum-test
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-7B](https://huggingface.co/Qwen/Qwen2.5-7B).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="chenggong1995/Qwen-2.5-Base-7B-gen8-math3to5_olympiads_aime-ghpo-cold0-hint50-prompt1-redonum-test", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/gongc1995-city-university-of-hong-kong/huggingface/runs/ogasbug2)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.16.0
|
| 38 |
+
- Transformers: 4.50.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.5.0
|
| 41 |
+
- Tokenizers: 0.21.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite GRPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{zhihong2024deepseekmath,
|
| 49 |
+
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
|
| 50 |
+
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2402.03300},
|
| 53 |
+
}
|
| 54 |
+
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
Cite TRL as:
|
| 58 |
+
|
| 59 |
+
```bibtex
|
| 60 |
+
@misc{vonwerra2022trl,
|
| 61 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 62 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 63 |
+
year = 2020,
|
| 64 |
+
journal = {GitHub repository},
|
| 65 |
+
publisher = {GitHub},
|
| 66 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 67 |
+
}
|
| 68 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.019440959097726123,
|
| 4 |
+
"train_runtime": 57675.0727,
|
| 5 |
+
"train_samples": 18328,
|
| 6 |
+
"train_samples_per_second": 0.318,
|
| 7 |
+
"train_steps_per_second": 0.003
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"eos_token_id": 151643,
|
| 4 |
+
"max_new_tokens": 2048,
|
| 5 |
+
"transformers_version": "4.50.0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.019440959097726123,
|
| 4 |
+
"train_runtime": 57675.0727,
|
| 5 |
+
"train_samples": 18328,
|
| 6 |
+
"train_samples_per_second": 0.318,
|
| 7 |
+
"train_steps_per_second": 0.003
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,512 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.9954198473282443,
|
| 6 |
+
"eval_steps": 200000,
|
| 7 |
+
"global_step": 163,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"clip_ratio": 0.0,
|
| 14 |
+
"completion_length": 567.619441986084,
|
| 15 |
+
"epoch": 0.0061068702290076335,
|
| 16 |
+
"grad_norm": 0.2135867174294605,
|
| 17 |
+
"learning_rate": 5.88235294117647e-08,
|
| 18 |
+
"loss": 0.0031,
|
| 19 |
+
"num_tokens": 743987.0,
|
| 20 |
+
"reward": 0.07700893143191934,
|
| 21 |
+
"reward_std": 0.1299287434667349,
|
| 22 |
+
"rewards/accuracy_reward": 0.06919642793945968,
|
| 23 |
+
"rewards/format_reward": 0.01562500058207661,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"clip_ratio": 0.0,
|
| 28 |
+
"completion_length": 588.1579504013062,
|
| 29 |
+
"epoch": 0.030534351145038167,
|
| 30 |
+
"grad_norm": 0.3343295200773406,
|
| 31 |
+
"learning_rate": 2.941176470588235e-07,
|
| 32 |
+
"loss": -0.0019,
|
| 33 |
+
"num_tokens": 3769305.0,
|
| 34 |
+
"reward": 0.0671037980064284,
|
| 35 |
+
"reward_std": 0.12649992539081722,
|
| 36 |
+
"rewards/accuracy_reward": 0.05803571452270262,
|
| 37 |
+
"rewards/format_reward": 0.018136161408619955,
|
| 38 |
+
"step": 5
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"clip_ratio": 0.0,
|
| 42 |
+
"completion_length": 588.1192253112793,
|
| 43 |
+
"epoch": 0.061068702290076333,
|
| 44 |
+
"grad_norm": 3.974587027805743,
|
| 45 |
+
"learning_rate": 5.88235294117647e-07,
|
| 46 |
+
"loss": 0.004,
|
| 47 |
+
"num_tokens": 7571663.0,
|
| 48 |
+
"reward": 0.07087053959257901,
|
| 49 |
+
"reward_std": 0.12776878643780948,
|
| 50 |
+
"rewards/accuracy_reward": 0.05870535783469677,
|
| 51 |
+
"rewards/format_reward": 0.02433035774156451,
|
| 52 |
+
"step": 10
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"clip_ratio": 0.0,
|
| 56 |
+
"completion_length": 606.0297157287598,
|
| 57 |
+
"epoch": 0.0916030534351145,
|
| 58 |
+
"grad_norm": 0.495391151906653,
|
| 59 |
+
"learning_rate": 8.823529411764705e-07,
|
| 60 |
+
"loss": 0.0015,
|
| 61 |
+
"num_tokens": 11421588.0,
|
| 62 |
+
"reward": 0.09140625372529029,
|
| 63 |
+
"reward_std": 0.1608576664701104,
|
| 64 |
+
"rewards/accuracy_reward": 0.06540178635623306,
|
| 65 |
+
"rewards/format_reward": 0.05200892963912338,
|
| 66 |
+
"step": 15
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"clip_ratio": 0.0,
|
| 70 |
+
"completion_length": 603.9748046875,
|
| 71 |
+
"epoch": 0.12213740458015267,
|
| 72 |
+
"grad_norm": 0.40463318371067986,
|
| 73 |
+
"learning_rate": 9.989585804326962e-07,
|
| 74 |
+
"loss": -0.0012,
|
| 75 |
+
"num_tokens": 15290931.0,
|
| 76 |
+
"reward": 0.20602679532021284,
|
| 77 |
+
"reward_std": 0.26069728564471006,
|
| 78 |
+
"rewards/accuracy_reward": 0.08125000051222742,
|
| 79 |
+
"rewards/format_reward": 0.24955357071012257,
|
| 80 |
+
"step": 20
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"clip_ratio": 0.0,
|
| 84 |
+
"completion_length": 637.4477951049805,
|
| 85 |
+
"epoch": 0.15267175572519084,
|
| 86 |
+
"grad_norm": 0.2621022081219828,
|
| 87 |
+
"learning_rate": 9.926100533780304e-07,
|
| 88 |
+
"loss": 0.0092,
|
| 89 |
+
"num_tokens": 19308993.0,
|
| 90 |
+
"reward": 0.4613839514553547,
|
| 91 |
+
"reward_std": 0.29145158380270003,
|
| 92 |
+
"rewards/accuracy_reward": 0.09285714323632419,
|
| 93 |
+
"rewards/format_reward": 0.737053570151329,
|
| 94 |
+
"step": 25
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"clip_ratio": 0.0,
|
| 98 |
+
"completion_length": 643.634407043457,
|
| 99 |
+
"epoch": 0.183206106870229,
|
| 100 |
+
"grad_norm": 0.6637031031614584,
|
| 101 |
+
"learning_rate": 9.805648919361503e-07,
|
| 102 |
+
"loss": 0.0201,
|
| 103 |
+
"num_tokens": 23260675.0,
|
| 104 |
+
"reward": 0.607031274586916,
|
| 105 |
+
"reward_std": 0.22586645130068064,
|
| 106 |
+
"rewards/accuracy_reward": 0.1455357144586742,
|
| 107 |
+
"rewards/format_reward": 0.9229910746216774,
|
| 108 |
+
"step": 30
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"clip_ratio": 0.0,
|
| 112 |
+
"completion_length": 719.7797180175781,
|
| 113 |
+
"epoch": 0.21374045801526717,
|
| 114 |
+
"grad_norm": 0.12398978901271052,
|
| 115 |
+
"learning_rate": 9.62962388596925e-07,
|
| 116 |
+
"loss": 0.0393,
|
| 117 |
+
"num_tokens": 27603904.0,
|
| 118 |
+
"reward": 0.6875000298023224,
|
| 119 |
+
"reward_std": 0.2122076230123639,
|
| 120 |
+
"rewards/accuracy_reward": 0.2053571424447,
|
| 121 |
+
"rewards/format_reward": 0.9642857119441033,
|
| 122 |
+
"step": 35
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"clip_ratio": 0.0,
|
| 126 |
+
"completion_length": 754.8406555175782,
|
| 127 |
+
"epoch": 0.24427480916030533,
|
| 128 |
+
"grad_norm": 0.11716820534931413,
|
| 129 |
+
"learning_rate": 9.400061019867678e-07,
|
| 130 |
+
"loss": 0.0414,
|
| 131 |
+
"num_tokens": 32059078.0,
|
| 132 |
+
"reward": 0.6993303894996643,
|
| 133 |
+
"reward_std": 0.18653424717485906,
|
| 134 |
+
"rewards/accuracy_reward": 0.21361607229337096,
|
| 135 |
+
"rewards/format_reward": 0.9714285716414451,
|
| 136 |
+
"step": 40
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"clip_ratio": 0.0,
|
| 140 |
+
"completion_length": 752.6609725952148,
|
| 141 |
+
"epoch": 0.2748091603053435,
|
| 142 |
+
"grad_norm": 0.1366862152370541,
|
| 143 |
+
"learning_rate": 9.11961502878777e-07,
|
| 144 |
+
"loss": 0.0295,
|
| 145 |
+
"num_tokens": 36473463.0,
|
| 146 |
+
"reward": 0.7407366394996643,
|
| 147 |
+
"reward_std": 0.16629975400865077,
|
| 148 |
+
"rewards/accuracy_reward": 0.24955357071012257,
|
| 149 |
+
"rewards/format_reward": 0.9823660656809807,
|
| 150 |
+
"step": 45
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"clip_ratio": 0.0,
|
| 154 |
+
"completion_length": 762.0527130126953,
|
| 155 |
+
"epoch": 0.3053435114503817,
|
| 156 |
+
"grad_norm": 0.11417326322107703,
|
| 157 |
+
"learning_rate": 8.791529042392812e-07,
|
| 158 |
+
"loss": 0.0308,
|
| 159 |
+
"num_tokens": 40938643.0,
|
| 160 |
+
"reward": 0.7233259305357933,
|
| 161 |
+
"reward_std": 0.16576984385028481,
|
| 162 |
+
"rewards/accuracy_reward": 0.23281249962747097,
|
| 163 |
+
"rewards/format_reward": 0.9810267806053161,
|
| 164 |
+
"step": 50
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"clip_ratio": 0.0,
|
| 168 |
+
"completion_length": 764.9165496826172,
|
| 169 |
+
"epoch": 0.33587786259541985,
|
| 170 |
+
"grad_norm": 0.10399028120391927,
|
| 171 |
+
"learning_rate": 8.419597108123053e-07,
|
| 172 |
+
"loss": 0.0269,
|
| 173 |
+
"num_tokens": 45439493.0,
|
| 174 |
+
"reward": 0.729910746216774,
|
| 175 |
+
"reward_std": 0.16336991311982274,
|
| 176 |
+
"rewards/accuracy_reward": 0.2368303578812629,
|
| 177 |
+
"rewards/format_reward": 0.9861607074737548,
|
| 178 |
+
"step": 55
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"clip_ratio": 0.0,
|
| 182 |
+
"completion_length": 750.571240234375,
|
| 183 |
+
"epoch": 0.366412213740458,
|
| 184 |
+
"grad_norm": 0.10084790893922295,
|
| 185 |
+
"learning_rate": 8.008120316124611e-07,
|
| 186 |
+
"loss": 0.0243,
|
| 187 |
+
"num_tokens": 49869428.0,
|
| 188 |
+
"reward": 0.7180803894996644,
|
| 189 |
+
"reward_std": 0.1501687964424491,
|
| 190 |
+
"rewards/accuracy_reward": 0.22611607052385807,
|
| 191 |
+
"rewards/format_reward": 0.9839285641908646,
|
| 192 |
+
"step": 60
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"clip_ratio": 0.0,
|
| 196 |
+
"completion_length": 788.064323425293,
|
| 197 |
+
"epoch": 0.3969465648854962,
|
| 198 |
+
"grad_norm": 0.0777401486625089,
|
| 199 |
+
"learning_rate": 7.561857060642119e-07,
|
| 200 |
+
"loss": 0.0195,
|
| 201 |
+
"num_tokens": 54497556.0,
|
| 202 |
+
"reward": 0.7145089581608772,
|
| 203 |
+
"reward_std": 0.16678392123430968,
|
| 204 |
+
"rewards/accuracy_reward": 0.22120535522699356,
|
| 205 |
+
"rewards/format_reward": 0.9866071388125419,
|
| 206 |
+
"step": 65
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"clip_ratio": 0.0,
|
| 210 |
+
"completion_length": 759.6870880126953,
|
| 211 |
+
"epoch": 0.42748091603053434,
|
| 212 |
+
"grad_norm": 0.08469355022666704,
|
| 213 |
+
"learning_rate": 7.085968013061584e-07,
|
| 214 |
+
"loss": 0.0277,
|
| 215 |
+
"num_tokens": 58963650.0,
|
| 216 |
+
"reward": 0.7053571730852127,
|
| 217 |
+
"reward_std": 0.13858543820679187,
|
| 218 |
+
"rewards/accuracy_reward": 0.21272321385331452,
|
| 219 |
+
"rewards/format_reward": 0.9852678492665291,
|
| 220 |
+
"step": 70
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"clip_ratio": 0.0,
|
| 224 |
+
"completion_length": 784.3569534301757,
|
| 225 |
+
"epoch": 0.4580152671755725,
|
| 226 |
+
"grad_norm": 0.09089115066810906,
|
| 227 |
+
"learning_rate": 6.585956442945531e-07,
|
| 228 |
+
"loss": 0.022,
|
| 229 |
+
"num_tokens": 63555393.0,
|
| 230 |
+
"reward": 0.7090402141213417,
|
| 231 |
+
"reward_std": 0.15061827255412935,
|
| 232 |
+
"rewards/accuracy_reward": 0.21651785587891936,
|
| 233 |
+
"rewards/format_reward": 0.9850446343421936,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"clip_ratio": 0.0,
|
| 238 |
+
"completion_length": 797.2703521728515,
|
| 239 |
+
"epoch": 0.48854961832061067,
|
| 240 |
+
"grad_norm": 0.09002554108722542,
|
| 241 |
+
"learning_rate": 6.06760457719898e-07,
|
| 242 |
+
"loss": 0.0238,
|
| 243 |
+
"num_tokens": 68216500.0,
|
| 244 |
+
"reward": 0.7666295006871223,
|
| 245 |
+
"reward_std": 0.16813623085618018,
|
| 246 |
+
"rewards/accuracy_reward": 0.27165178642608223,
|
| 247 |
+
"rewards/format_reward": 0.9899553492665291,
|
| 248 |
+
"step": 80
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"clip_ratio": 0.0,
|
| 252 |
+
"completion_length": 761.4013763427735,
|
| 253 |
+
"epoch": 0.5190839694656488,
|
| 254 |
+
"grad_norm": 0.08498851686715876,
|
| 255 |
+
"learning_rate": 5.536906733320815e-07,
|
| 256 |
+
"loss": 0.0158,
|
| 257 |
+
"num_tokens": 72721850.0,
|
| 258 |
+
"reward": 0.7462053880095482,
|
| 259 |
+
"reward_std": 0.1457503356039524,
|
| 260 |
+
"rewards/accuracy_reward": 0.2502232126891613,
|
| 261 |
+
"rewards/format_reward": 0.9919642806053162,
|
| 262 |
+
"step": 85
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"clip_ratio": 0.0,
|
| 266 |
+
"completion_length": 766.9694580078125,
|
| 267 |
+
"epoch": 0.549618320610687,
|
| 268 |
+
"grad_norm": 0.07561608030311687,
|
| 269 |
+
"learning_rate": 5e-07,
|
| 270 |
+
"loss": 0.024,
|
| 271 |
+
"num_tokens": 77220857.0,
|
| 272 |
+
"reward": 0.739285746216774,
|
| 273 |
+
"reward_std": 0.14479873944073915,
|
| 274 |
+
"rewards/accuracy_reward": 0.2448660720139742,
|
| 275 |
+
"rewards/format_reward": 0.9888392820954323,
|
| 276 |
+
"step": 90
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"clip_ratio": 0.0,
|
| 280 |
+
"completion_length": 760.7444534301758,
|
| 281 |
+
"epoch": 0.5801526717557252,
|
| 282 |
+
"grad_norm": 0.08175108965516903,
|
| 283 |
+
"learning_rate": 4.463093266679185e-07,
|
| 284 |
+
"loss": 0.0241,
|
| 285 |
+
"num_tokens": 81689720.0,
|
| 286 |
+
"reward": 0.7234375298023223,
|
| 287 |
+
"reward_std": 0.1551271199248731,
|
| 288 |
+
"rewards/accuracy_reward": 0.22790178465656935,
|
| 289 |
+
"rewards/format_reward": 0.9910714194178581,
|
| 290 |
+
"step": 95
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"clip_ratio": 0.0,
|
| 294 |
+
"completion_length": 759.4685607910156,
|
| 295 |
+
"epoch": 0.6106870229007634,
|
| 296 |
+
"grad_norm": 0.09009365686861091,
|
| 297 |
+
"learning_rate": 3.932395422801019e-07,
|
| 298 |
+
"loss": 0.0283,
|
| 299 |
+
"num_tokens": 86136451.0,
|
| 300 |
+
"reward": 0.7452009245753288,
|
| 301 |
+
"reward_std": 0.1618154514580965,
|
| 302 |
+
"rewards/accuracy_reward": 0.2524553562514484,
|
| 303 |
+
"rewards/format_reward": 0.9854910641908645,
|
| 304 |
+
"step": 100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"clip_ratio": 0.0,
|
| 308 |
+
"completion_length": 739.8957916259766,
|
| 309 |
+
"epoch": 0.6412213740458015,
|
| 310 |
+
"grad_norm": 0.09858205620896988,
|
| 311 |
+
"learning_rate": 3.41404355705447e-07,
|
| 312 |
+
"loss": 0.0175,
|
| 313 |
+
"num_tokens": 90456928.0,
|
| 314 |
+
"reward": 0.7717634305357933,
|
| 315 |
+
"reward_std": 0.1629006579518318,
|
| 316 |
+
"rewards/accuracy_reward": 0.2754464283585548,
|
| 317 |
+
"rewards/format_reward": 0.9926339209079742,
|
| 318 |
+
"step": 105
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"clip_ratio": 0.0,
|
| 322 |
+
"completion_length": 779.8815078735352,
|
| 323 |
+
"epoch": 0.6717557251908397,
|
| 324 |
+
"grad_norm": 0.08118289189251325,
|
| 325 |
+
"learning_rate": 2.914031986938417e-07,
|
| 326 |
+
"loss": 0.019,
|
| 327 |
+
"num_tokens": 95039765.0,
|
| 328 |
+
"reward": 0.7184152096509934,
|
| 329 |
+
"reward_std": 0.15354990400373936,
|
| 330 |
+
"rewards/accuracy_reward": 0.2238839288474992,
|
| 331 |
+
"rewards/format_reward": 0.9890624955296516,
|
| 332 |
+
"step": 110
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"clip_ratio": 0.0,
|
| 336 |
+
"completion_length": 743.4281570434571,
|
| 337 |
+
"epoch": 0.7022900763358778,
|
| 338 |
+
"grad_norm": 0.08942937354761228,
|
| 339 |
+
"learning_rate": 2.4381429393578815e-07,
|
| 340 |
+
"loss": 0.0134,
|
| 341 |
+
"num_tokens": 99422555.0,
|
| 342 |
+
"reward": 0.7406250327825546,
|
| 343 |
+
"reward_std": 0.15929818488657474,
|
| 344 |
+
"rewards/accuracy_reward": 0.24397321371361613,
|
| 345 |
+
"rewards/format_reward": 0.9933035641908645,
|
| 346 |
+
"step": 115
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"clip_ratio": 0.0,
|
| 350 |
+
"completion_length": 755.4859725952149,
|
| 351 |
+
"epoch": 0.732824427480916,
|
| 352 |
+
"grad_norm": 0.08614513315269835,
|
| 353 |
+
"learning_rate": 1.991879683875386e-07,
|
| 354 |
+
"loss": 0.0216,
|
| 355 |
+
"num_tokens": 103884636.0,
|
| 356 |
+
"reward": 0.7185268118977547,
|
| 357 |
+
"reward_std": 0.14079238111153244,
|
| 358 |
+
"rewards/accuracy_reward": 0.22388392696157097,
|
| 359 |
+
"rewards/format_reward": 0.9892857059836387,
|
| 360 |
+
"step": 120
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"clip_ratio": 0.0,
|
| 364 |
+
"completion_length": 738.5589630126954,
|
| 365 |
+
"epoch": 0.7633587786259542,
|
| 366 |
+
"grad_norm": 0.0987941233877778,
|
| 367 |
+
"learning_rate": 1.5804028918769485e-07,
|
| 368 |
+
"loss": 0.0177,
|
| 369 |
+
"num_tokens": 108269772.0,
|
| 370 |
+
"reward": 0.7502232447266579,
|
| 371 |
+
"reward_std": 0.1488088957965374,
|
| 372 |
+
"rewards/accuracy_reward": 0.2546874986961484,
|
| 373 |
+
"rewards/format_reward": 0.9910714223980903,
|
| 374 |
+
"step": 125
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"clip_ratio": 0.0,
|
| 378 |
+
"completion_length": 718.1944564819336,
|
| 379 |
+
"epoch": 0.7938931297709924,
|
| 380 |
+
"grad_norm": 0.08957435241917934,
|
| 381 |
+
"learning_rate": 1.2084709576071883e-07,
|
| 382 |
+
"loss": 0.0164,
|
| 383 |
+
"num_tokens": 112486547.0,
|
| 384 |
+
"reward": 0.7435268193483353,
|
| 385 |
+
"reward_std": 0.15864104311913252,
|
| 386 |
+
"rewards/accuracy_reward": 0.24799107126891612,
|
| 387 |
+
"rewards/format_reward": 0.9910714194178581,
|
| 388 |
+
"step": 130
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"clip_ratio": 0.0,
|
| 392 |
+
"completion_length": 771.4886535644531,
|
| 393 |
+
"epoch": 0.8244274809160306,
|
| 394 |
+
"grad_norm": 0.10033485818985768,
|
| 395 |
+
"learning_rate": 8.803849712122291e-08,
|
| 396 |
+
"loss": 0.0191,
|
| 397 |
+
"num_tokens": 117017984.0,
|
| 398 |
+
"reward": 0.7385044932365418,
|
| 399 |
+
"reward_std": 0.15429062955081463,
|
| 400 |
+
"rewards/accuracy_reward": 0.24508928610011935,
|
| 401 |
+
"rewards/format_reward": 0.9868303492665291,
|
| 402 |
+
"step": 135
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"clip_ratio": 0.0,
|
| 406 |
+
"completion_length": 753.3828475952148,
|
| 407 |
+
"epoch": 0.8549618320610687,
|
| 408 |
+
"grad_norm": 0.08591161076983043,
|
| 409 |
+
"learning_rate": 5.999389801323218e-08,
|
| 410 |
+
"loss": 0.017,
|
| 411 |
+
"num_tokens": 121422939.0,
|
| 412 |
+
"reward": 0.7706473574042321,
|
| 413 |
+
"reward_std": 0.16626517940312624,
|
| 414 |
+
"rewards/accuracy_reward": 0.2743303582072258,
|
| 415 |
+
"rewards/format_reward": 0.9926339253783226,
|
| 416 |
+
"step": 140
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"clip_ratio": 0.0,
|
| 420 |
+
"completion_length": 737.1861953735352,
|
| 421 |
+
"epoch": 0.8854961832061069,
|
| 422 |
+
"grad_norm": 0.08133023369347758,
|
| 423 |
+
"learning_rate": 3.7037611403075096e-08,
|
| 424 |
+
"loss": 0.0211,
|
| 425 |
+
"num_tokens": 125792325.0,
|
| 426 |
+
"reward": 0.7737723588943481,
|
| 427 |
+
"reward_std": 0.14874637452885509,
|
| 428 |
+
"rewards/accuracy_reward": 0.2774553569033742,
|
| 429 |
+
"rewards/format_reward": 0.9926339238882065,
|
| 430 |
+
"step": 145
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"clip_ratio": 0.0,
|
| 434 |
+
"completion_length": 719.9216812133789,
|
| 435 |
+
"epoch": 0.916030534351145,
|
| 436 |
+
"grad_norm": 0.09062176860429237,
|
| 437 |
+
"learning_rate": 1.943510806384968e-08,
|
| 438 |
+
"loss": 0.0223,
|
| 439 |
+
"num_tokens": 130100454.0,
|
| 440 |
+
"reward": 0.7459821745753288,
|
| 441 |
+
"reward_std": 0.1444489900022745,
|
| 442 |
+
"rewards/accuracy_reward": 0.24977678433060646,
|
| 443 |
+
"rewards/format_reward": 0.9924107134342194,
|
| 444 |
+
"step": 150
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"clip_ratio": 0.0,
|
| 448 |
+
"completion_length": 725.9875335693359,
|
| 449 |
+
"epoch": 0.9465648854961832,
|
| 450 |
+
"grad_norm": 0.0910938816922007,
|
| 451 |
+
"learning_rate": 7.389946621969678e-09,
|
| 452 |
+
"loss": 0.0111,
|
| 453 |
+
"num_tokens": 134418750.0,
|
| 454 |
+
"reward": 0.7150669932365418,
|
| 455 |
+
"reward_std": 0.145760334469378,
|
| 456 |
+
"rewards/accuracy_reward": 0.21897321371361614,
|
| 457 |
+
"rewards/format_reward": 0.9921874910593033,
|
| 458 |
+
"step": 155
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"clip_ratio": 0.0,
|
| 462 |
+
"completion_length": 729.7480224609375,
|
| 463 |
+
"epoch": 0.9770992366412213,
|
| 464 |
+
"grad_norm": 0.0976254681849543,
|
| 465 |
+
"learning_rate": 1.0414195673039138e-09,
|
| 466 |
+
"loss": 0.0201,
|
| 467 |
+
"num_tokens": 138738941.0,
|
| 468 |
+
"reward": 0.7399553954601288,
|
| 469 |
+
"reward_std": 0.13362135970965028,
|
| 470 |
+
"rewards/accuracy_reward": 0.24397321399301292,
|
| 471 |
+
"rewards/format_reward": 0.9919642806053162,
|
| 472 |
+
"step": 160
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"clip_ratio": 0.0,
|
| 476 |
+
"completion_length": 757.2314198811849,
|
| 477 |
+
"epoch": 0.9954198473282443,
|
| 478 |
+
"num_tokens": 141385215.0,
|
| 479 |
+
"reward": 0.7395833656191826,
|
| 480 |
+
"reward_std": 0.1369063208500544,
|
| 481 |
+
"rewards/accuracy_reward": 0.2447916651920726,
|
| 482 |
+
"rewards/format_reward": 0.9895833333333334,
|
| 483 |
+
"step": 163,
|
| 484 |
+
"total_flos": 0.0,
|
| 485 |
+
"train_loss": 0.019440959097726123,
|
| 486 |
+
"train_runtime": 57675.0727,
|
| 487 |
+
"train_samples_per_second": 0.318,
|
| 488 |
+
"train_steps_per_second": 0.003
|
| 489 |
+
}
|
| 490 |
+
],
|
| 491 |
+
"logging_steps": 5,
|
| 492 |
+
"max_steps": 163,
|
| 493 |
+
"num_input_tokens_seen": 0,
|
| 494 |
+
"num_train_epochs": 1,
|
| 495 |
+
"save_steps": 500,
|
| 496 |
+
"stateful_callbacks": {
|
| 497 |
+
"TrainerControl": {
|
| 498 |
+
"args": {
|
| 499 |
+
"should_epoch_stop": false,
|
| 500 |
+
"should_evaluate": false,
|
| 501 |
+
"should_log": false,
|
| 502 |
+
"should_save": true,
|
| 503 |
+
"should_training_stop": true
|
| 504 |
+
},
|
| 505 |
+
"attributes": {}
|
| 506 |
+
}
|
| 507 |
+
},
|
| 508 |
+
"total_flos": 0.0,
|
| 509 |
+
"train_batch_size": 16,
|
| 510 |
+
"trial_name": null,
|
| 511 |
+
"trial_params": null
|
| 512 |
+
}
|