🚀 Llama-3-OffsetBias-RM-8B模型卡片
Llama-3-OffsetBias-RM-8B 是一个在OffsetBias数据集上训练的奖励模型。它经过训练,在评估模型中常见的各种评估偏差方面表现得更加稳健。该模型在论文 OffsetBias: Leveraging Debiased Data for Tuning Evaluators 中被提出。
🚀 快速开始
直接使用
from transformers import AutoTokenizer, pipeline
import torch
model_name = "NCSOFT/Llama-3-OffsetBias-RM-8B"
rm_tokenizer = AutoTokenizer.from_pretrained(model_name)
rm_pipe = pipeline(
"sentiment-analysis",
model=model_name,
device="auto",
tokenizer=rm_tokenizer,
model_kwargs={"torch_dtype": torch.bfloat16}
)
pipe_kwargs = {
"return_all_scores": True,
"function_to_apply": "none",
"batch_size": 1
}
chat = [
{"role": "user", "content": "Hello, how are you?"},
{"role": "assistant", "content": "I'm doing great. How can I help you today?"},
{"role": "user", "content": "I'd like to show off how chat templating works!"},
]
test_texts = [rm_tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=False).replace(rm_tokenizer.bos_token, "")]
pipe_outputs = rm_pipe(test_texts, **pipe_kwargs)
rewards = [output[0]["score"] for output in pipe_outputs]
✨ 主要特性
📚 详细文档
模型详情
模型描述
Llama-3-OffsetBias-RM-8B 使用 sfairXC/FsfairX-LLaMA3-RM-v0.1 作为基础模型,它基于Meta Llama 3构建。一个中间奖励模型是从 Llama-3-8B-Instruct 开始训练的,使用了 FsfairX-LLaMA3-RM 模型训练所用数据集的一个子集,并结合了 NCSOFT/offsetbias 数据集。然后将这个中间模型与 FsfairX-LLaMA3-RM 模型合并,创建了 Llama-3-OffsetBias-RM-8B。
模型来源
评估
RewardBench结果
指标 |
得分 |
Chat |
97.21 |
Chat Hard |
80.70 |
Safety |
89.01 |
Reasoning |
90.60 |
EvalBiasBench结果
指标 |
得分 |
Length |
82.4 |
Concreteness |
92.9 |
Empty Reference |
46.2 |
Content Continuation |
100.0 |
Nested Instruction |
83.3 |
Familiar Knowledge |
58.3 |
📄 许可证
该模型使用的许可证为 META LLAMA 3 COMMUNITY LICENSE AGREEMENT。
🔖 引用
@misc{park2024offsetbias,
title={OffsetBias: Leveraging Debiased Data for Tuning Evaluators},
author={Junsoo Park and Seungyeon Jwa and Meiying Ren and Daeyoung Kim and Sanghyuk Choi},
year={2024},
eprint={2407.06551},
archivePrefix={arXiv},
primaryClass={cs.CL}
}