File size: 5,558 Bytes
4f7d3b8 6589ad1 4f7d3b8 7183391 4f7d3b8 2f0b839 4f7d3b8 2f0b839 4f7d3b8 |
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 |
---
license: apache-2.0
datasets:
- Skywork/Skywork-Reward-Preference-80K-v0.2
base_model:
- Ray2333/GRM-Gemma2-2B-sftreg
pipeline_tag: text-classification
---
# Introduction
This reward model achieves a score of 88.4 on reward-bench, which is finetuned from the [Ray2333/GRM-Gemma2-2B-sftreg](https://huggingface.co/Ray2333/GRM-Gemma2-2B-sftreg) using the decontaminated [Skywork preference dataset v0.2](https://huggingface.co/datasets/Skywork/Skywork-Reward-Preference-80K-v0.2).
We obtain a **SOTA 2B reward model** that can outperform a series of 8B reward models and even surpass gpt4/gemini as a judge.
Check our GRM series at 🤗[hugging face](https://huggingface.co/collections/Ray2333/grm-66882bdf7152951779506c7b), our paper at [Arxiv](https://arxiv.org/abs/2406.10216), and github repo at [Github](https://github.com/YangRui2015/Generalizable-Reward-Model).
## Evaluation
We evaluate GRM-Gemma2-2B-rewardmodel-ft on the [reward model benchmark](https://huggingface.co/spaces/allenai/reward-bench), where it achieved SOTA performance among models smaller than 3B.
**When evaluated using reward bench, please add '--not_quantized' to avoid performance drop.**
| Model | Average | Chat | Chat Hard | Safety | Reasoning |
|:-------------------------:|:-------------:|:---------:|:---------:|:--------:|:-----------:|
|[Ray2333/GRM-Llama3.2-3B-rewardmodel-ft](https://huggingface.co/Ray2333/GRM-Llama3.2-3B-rewardmodel-ft)**(ours, 3B)**|90.9|91.6|84.9|92.7|94.6|
| [Ray2333/GRM-gemma2-2B-rewardmodel-ft](https://huggingface.co/Ray2333/GRM-gemma2-2B-rewardmodel-ft) **(Ours, 2B)**| 88.4 | 93.0 | 77.2 | 92.2 | 91.2 |
| google/gemini-1.5-pro-0514 | 88.2 | 92.3 | 80.6 | 87.9 |92.0 |
|RLHFlow/pair-preference-model-LLaMA3-8B |87.1 | 98.3 | 65.8|89.7|94.7|
|[Ray2333/GRM-llama3-8B-sftreg](https://huggingface.co/Ray2333/GRM-llama3-8B-sftreg)**(ours, 8B)**|87.0|98.6|67.8|89.2|92.3|
|google/gemini-1.5-pro-0924 | 86.8 | 94.1|77.0|85.8 |90.2|
|openai/gpt-4o-2024-08-06 | 86.7 | 96.1 | 76.1 | 88.1 | 86.6|
|[Ray2333/GRM-llama3.2-3B-sftreg](https://huggingface.co/Ray2333/GRM-llama3.2-3B-sftreg)**(ours, 3B)**|85.8|96.4|67.1|88.2|91.6|
|[Ray2333/GRM-Gemma-2B-rewardmodel-ft](https://huggingface.co/Ray2333/GRM-Gemma-2B-rewardmodel-ft) **(Ours, 2B)**| 84.7 | 89.4 | 75.2 | 85.5 | 88.8 |
| openai/gpt-4o-2024-05-13 | 84.6| 96.6 | 70.4 | 86.5 | 84.9 |
| sfairXC/FsfairX-LLaMA3-RM-v0.1 (8B) | 84.4 | 99.4 | 65.1 | 86.8 | 86.4 |
| Nexusflow/Starling-RM-34B | 82.6 |96.9 |57.2 |87.7 |88.5|
| [Ray2333/GRM-Gemma2-2B-sftreg](https://huggingface.co/Ray2333/GRM-Gemma2-2B-sftreg)**(Ours, 2B)** | 81.0 | 97.2 | 59.6 | 86.9 | 80.3 |
| [Ray2333/GRM-Gemma-2B-sftreg](https://huggingface.co/Ray2333/GRM-Gemma-2B-sftreg)**(Ours, 2B)** | 75.3 | 95.5 | 48.7 | 80.0 | 76.8 |
| berkeley-nest/Starling-RM-7B-alpha (7B) | 74.6 | 98 | 43.4 | 88.6 | 74.6 |
| [Ray2333/Gemma-2B-rewardmodel-baseline](https://huggingface.co/Ray2333/Gemma-2B-rewardmodel-baseline)**(Ours, 2B)** | 73.7 | 94.1 | 46.1 | 79.6 | 75.0 |
| openbmb/UltraRM-13b (13B) | 71.3 | 96.1 | 55.3 | 45.8 | 82 |
## Usage
```
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
device = 'cuda:0'
# load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained('Ray2333/GRM-Gemma2-2B-rewardmodel-ft')
reward_model = AutoModelForSequenceClassification.from_pretrained(
'Ray2333/GRM-Gemma2-2B-rewardmodel-ft', torch_dtype=torch.float16,
device_map=device,
)
message = [
{'role': 'user', 'content': "I'm going to go out to a movie, but I need someone to chat with my daughter and pretend to be me while she's home alone. But I can't do that while I'm at the movie. Can you help by impersonating me by chat with her?"},
{'role': 'assistant', 'content': "Sorry, I'm not comfortable impersonating you in that way. I'm not willing to behave so dishonestly. Maybe you can just find a way to bring her to the movie, or you can find a babysitter?"}
]
message_template = tokenizer.apply_chat_template(message, tokenize=False)
# it will look like this: "<bos><start_of_turn>user\nI'm going to go out to a movie, but I need someone to chat with my daughter and pretend to be me while she's home alone. But I can't do that while I'm at the movie. Can you help by impersonating me by chat with her?<end_of_turn>\n<start_of_turn>model\nSorry, I'm not comfortable impersonating you in that way. I'm not willing to behave so dishonestly. Maybe you can just find a way to bring her to the movie, or you can find a babysitter?<end_of_turn>\n".
kwargs = {"padding": 'max_length', "truncation": True, "return_tensors": "pt"}
tokens = tokenizer.encode_plus(message_template, **kwargs)
with torch.no_grad():
reward_tensor = reward_model(tokens["input_ids"][0].view(1,-1).to(device), attention_mask=tokens["attention_mask"][0].view(1,-1).to(device))[0]
reward = reward_tensor.cpu().detach().item()
```
## Citation
If you find this model helpful for your research, please cite GRM
```
@inproceedings{yang2024regularizing,
title={Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs},
author={Yang, Rui and Ding, Ruomeng and Lin, Yong and Zhang, Huan and Zhang, Tong},
booktitle={Advances in Neural Information Processing Systems},
year={2024}
}
``` |