Hugging Face's logo

sky-2002
/
SmolLM2-360M-GRPO-v0

Transformers
Safetensors
Generated from Trainer
trl
grpo
Model card Files
xet
 Community
SmolLM2-360M-GRPO-v0
File size: 2,587 Bytes
18c1e88
 
 
 
 
 
 
 
 
 
 
 
 
 
eb2403a
18c1e88
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eb2403a
 
 
 
 
 
 
 
 
18c1e88
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
---
base_model: HuggingFaceTB/SmolLM2-360M
library_name: transformers
model_name: SmolLM2-360M-GRPO-v0
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---

# Model Card for SmolLM2-360M-GRPO-v0

This model is a fine-tuned version of [HuggingFaceTB/SmolLM2-360M](https://huggingface.co/HuggingFaceTB/SmolLM2-360M).
It has been finetuned using [TRL](https://github.com/huggingface/trl).

## Quick start

```python
from transformers import pipeline

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?"
generator = pipeline("text-generation", model="sky-2002/SmolLM2-360M-GRPO-v0", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```

## Training procedure

[<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/aathatte2002-indian-institute-of-technology/SmolLM-135M-finetune/runs/szfjiiio) 


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) and using the `lamini/taylor_swift` dataset.

## Evals
Referring this [blog post](https://datawizz.ai/blog/grpo-fine-tuning-qwen-0-5b-vs-openai-o1-preview), used a similar evaluation method:

| Model | Average ROUGE-L |
|-------|-----------------|
| Qwen-0.5B | 0.3313 |
| SmolLM2-360M-GRPO-v0 | 0.1644 |

### Framework versions

- TRL: 0.15.0.dev0
- Transformers: 4.49.0.dev0
- Pytorch: 2.5.1
- Datasets: 3.2.0
- Tokenizers: 0.21.0

## Citations

Cite GRPO as:

```bibtex
@article{zhihong2024deepseekmath,
    title        = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
    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},
    year         = 2024,
    eprint       = {arXiv:2402.03300},
}

```

Cite TRL as:
    
```bibtex
@misc{vonwerra2022trl,
	title        = {{TRL: Transformer Reinforcement Learning}},
	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},
	year         = 2020,
	journal      = {GitHub repository},
	publisher    = {GitHub},
	howpublished = {\url{https://github.com/huggingface/trl}}
}
```