GRM Llama3.1 8B Rewardmodel Ft

🚀 GRM-Llama3.1_8B_rewardmodel-ft獎勵模型

本獎勵模型基於GRM-Llama3.1-8B-sftreg模型，使用經過淨化處理的Skywork偏好數據集v0.2進行微調，在reward-bench上取得了92.6的評分。該模型可用於文本分類任務，為相關研究和應用提供了有力支持。

🚀 快速開始

本獎勵模型基於GRM-Llama3.1-8B-sftreg模型微調而來，使用了經過淨化處理的Skywork偏好數據集v0.2。你可以通過以下鏈接查看我們的GRM系列模型、相關論文以及GitHub代碼倉庫：

• GRM系列模型：🤗hugging face
• 相關論文：Arxiv
• GitHub代碼倉庫：Github

✨ 主要特性

• 高性能：在reward-bench上取得了92.6的評分，展現出優秀的性能。
• 數據優質：使用經過淨化處理的Skywork偏好數據集v0.2進行微調，數據質量有保障。

📦 安裝指南

文檔中未提及具體安裝步驟，故跳過該章節。

💻 使用示例

基礎用法

import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification

device = 'cuda:0'
# load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained('Ray2333/GRM-Llama3.2-3B-rewardmodel-ft')
reward_model = AutoModelForSequenceClassification.from_pretrained(
                'Ray2333/GRM-Llama3.2-3B-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": 'longest', "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()

📚 詳細文檔

評估

我們在獎勵模型基準測試上對GRM_Llama3.1_8B_rewardmodel-ft進行了評估。

⚠️ 重要提示

當使用reward bench進行評估時，請添加 '--not_quantized' 以避免性能下降。

🔧 技術細節

文檔中未提及具體技術細節，故跳過該章節。

📄 許可證

本項目採用Apache 2.0許可證。

📚 引用

如果你發現該模型對你的研究有幫助，請引用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}
}