license: mit
datasets:
- weqweasdas/preference_dataset_mixture2_and_safe_pku
pipeline_tag: text-classification
Introduction
The Generalizable Reward Model (GRM) aims to enhance the generalization ability of reward models for LLMs through regularizing the hidden states.
Paper: Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs.
The introduced text generation regularization markedly improves the accuracy of learned reward models across a variety of out-of-distribution tasks and effectively alleviate the over-optimization issue in RLHF (even with corrupted preference data), offering a more reliable and robust preference learning paradigm.
This reward model is finetuned from gemma-2b-it using the weqweasdas/preference_dataset_mixture2_and_safe_pku dataset.
Evaluation
We evaluate GRM 2B on the reward model benchmark, which achieves the SOTA 2B Bradley–Terry model Performance.
Model | Average | Chat | Chat Hard | Safety | Reasoning |
---|---|---|---|---|---|
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(Ours, 2B) | 73.7 | 94.1 | 46.1 | 79.6 | 75.0 |
stabilityai/stablelm-zephyr-3b (3B) | 73.1 | 86.3 | 60.1 | 70.3 | 75.7 |
openbmb/UltraRM-13b (13B) | 71.3 | 96.1 | 55.3 | 45.8 | 82 |
Usage
Note: Please download the model.py
file from this repository to ensure the structure is loaded correctly and verify that the v_head
is properly initialized.
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
device = 'cuda:2'
# load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained('Ray2333/GRM-Gemma-2B-sftreg')
reward_model = AutoModelForSequenceClassification.from_pretrained(
'Ray2333/GRM-Gemma-2B-sftreg', torch_dtype=torch.float16, trust_remote_code=True,
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))
reward = reward_tensor.cpu().detach().item()
Citation
If you find this model helpful for your research, please cite GRM
@article{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},
journal={arXiv preprint arXiv:2406.10216},
year={2024}
}