File size: 4,798 Bytes
43ecab1 de93377 43ecab1 |
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 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 |
\---
license: apache-2.0
\---
### Storm-7B
> **Developed by**: [Jie Liu](https://jieliu.site/)$^{*1,2}$, [Zhanhui Zhou](https://scholar.google.com/citations?user=SbACfYQAAAAJ&hl=zh-CN)$^{*2}$, [Chao Yang](https://scholar.google.com/citations?user=5KRbHPMAAAAJ&hl=zh-CN)$^{2}$, [Han-Sen Zhong](https://scholar.google.com.hk/citations?user=X_ZfX8sAAAAJ&hl=zh-CN)$^{2}$, and [Wanli Ouyang](https://wlouyang.github.io/)$^{1,2}$.
>
> $^{1}$MMLab, The Chinese University of Hong Kong $^{2}$Shanghai AI Laboratory
#### Introduction
We released Storm-7B, the first open-source language model comparable to the GPT-4 series on the [AlpacaEval 2.0](https://tatsu-lab.github.io/alpaca_eval/) leaderboard, ranking 3rd in length-controlled win rate.
The recipe for this model is simple: 1) fine-tuning from [Openchat-3.5-0106](https://huggingface.co/openchat/openchat-3.5-0106), 2) applying iterative DPO training, a variant of DPO where a language model iteratively learns from the preferences of the trained reward model. We will release our technical report and code as soon as possible.
A snapshot of the AlpacaEval 2.0 leaderboard (2024/4/28) is listed below:
| | **LC Win Rate** | **Win Rate** |
| :----------------------: | :-------------: | :----------: |
| GPT-4 Turbo (04/09) | 55.0% | 46.1% |
| GPT-4 Preview (11/06) | 50.0% | 50.0% |
| **Storm-7B** | 48.9% | 52.5% |
| Nanbeige Plus Chat v0.1 | 44.5% | 56.7% |
| Qwen1.5 110B Chat | 43.9% | 33.8% |
| Aligner 2B+Claude 3 Opus | 41.8% | 34.5% |
| Claude 3 Opus (02/29) | 40.5% | 29.1% |
| GPT-4 | 38.1% | 23.6% |
| openchat-3.5-0106 | 15.4% | 10.1% |
Please refer to the [leaderboard webpage](https://tatsu-lab.github.io/alpaca_eval/) for up-to-date results.
We also conducted preliminary evaluations on other benchmarks and observed no significant degradation.
| | ARC | HellaSwag | MMLU | TruthfulQA | Winogrande | Avg. |
| ----------------- | ----- | --------- | ----- | ---------- | ---------- | ----- |
| **Storm-7B** | 67.58 | 80.97 | 62.21 | 57.24 | 80.51 | 69.70 |
| openchat-3.5-0106 | 66.38 | 83.00 | 63.47 | 52.55 | 81.06 | 69.29 |
| internlm2-7b | 58.02 | 81.24 | 65.24 | 48.73 | 83.82 | 67.41 |
| gemma-7B | 61.09 | 82.20 | 64.56 | 44.79 | 79.01 | 66.33 |
| Yi-9B | 61.18 | 78.82 | 70.06 | 42.45 | 77.51 | 66.00 |
| Meta-Llama-3-8B | 59.47 | 82.09 | 66.69 | 43.90 | 77.35 | 65.90 |
| Mistral-7B-v0.1 | 59.98 | 83.31 | 64.16 | 42.15 | 78.37 | 65.59 |
| Qwen-7b | 51.37 | 78.47 | 59.84 | 47.79 | 72.69 | 62.03 |
#### Uses
Our model uses the same chat template as [Openchat-3.5-0106](https://huggingface.co/openchat/openchat-3.5-0106). A sample code snippet for inference using our model is provided below.
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda"
model = AutoModelForCausalLM.from_pretrained("jieliu/Storm-7B").to(device)
tokenizer = AutoTokenizer.from_pretrained("jieliu/Storm-7B")
model.eval().requires_grad_(False)
def generate_response(prompt):
input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to(device)
outputs = model.generate(
input_ids,
max_length=2048,
do_sample=True,
temperature=1.0,
pad_token_id=tokenizer.pad_token_id,
eos_token_id=tokenizer.eos_token_id,
)
response_ids = outputs[0]
response_text = tokenizer.decode(response_ids, skip_special_tokens=True)
return response_text
prompt = "I'm trying to teach myself to have nicer handwriting. Can you help?"
input_prompt = f"GPT4 Correct User: {prompt}<|end_of_turn|>GPT4 Correct Assistant:"
response_text = generate_response(input_prompt)
print("Response:", response_text)
```
#### Limitations
Storm-7B is a quick demonstration that a language model, fine-tuned with AI feedback, can easily surpass or match state-of-the-art models, as assessed by the same AI feedback. However, this improvement on the automatic leaderboard may not necessarily indicate better alignment with human intentions. Our model therefore represents a critical, preliminary reevaluation of the RLAIF paradigm, questioning how much learning from and being evaluated by AI feedback aligns with actual human preferences.
#### Citation
```
@misc{liu2024storm,
title = {Storm-7B},
url = {},
author = {Jie Liu and Zhanhui Zhou and Chao Yang and Han-Sen Zhong and Wanli Ouyang},
month = {April},
year = {2024}
}
```
|