Felladrin/ChatML-openhermes2.5-dpo-binarized-alpha
收藏Hugging Face2024-02-10 更新2024-03-04 收录
下载链接：
https://hf-mirror.com/datasets/Felladrin/ChatML-openhermes2.5-dpo-binarized-alpha
下载链接
链接失效反馈官方服务：
资源简介：
---
language:
- en
size_categories:
- 1K<n<10K
---
[argilla/OpenHermes2.5-dpo-binarized-alpha](https://huggingface.co/datasets/argilla/OpenHermes2.5-dpo-binarized-alpha) in ChatML format, ready to use in [HuggingFace TRL's DPO Trainer](https://huggingface.co/docs/trl/main/en/dpo_trainer).
Python code used for conversion:
```python
from datasets import load_dataset
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("Felladrin/Llama-160M-Chat-v1")
dataset = load_dataset("argilla/openhermes2.5-dpo-binarized-alpha", split="train")
def format(columns):
return {
"prompt": tokenizer.apply_chat_template(columns["chosen"][:-1], tokenize=False, add_generation_prompt=True),
"chosen": f"{columns['chosen'][-1]['content']}<|im_end|>",
"rejected": f"{columns['rejected'][-1]['content']}<|im_end|>",
}
dataset.map(format).select_columns(['prompt', 'chosen', 'rejected', 'category', 'source', 'chosen_model', 'rejected_model', 'rejected_score', 'chosen_score']).to_parquet("train.parquet")
```
提供机构：
Felladrin
原始信息汇总
数据集概述
数据集名称
- 名称: argilla/OpenHermes2.5-dpo-binarized-alpha
数据集格式
- 格式: ChatML
数据集大小
- 大小: 1K<n<10K
数据集用途
- 用途: 适用于HuggingFace TRLs DPO Trainer
数据转换代码
-
Python代码: python from datasets import load_dataset from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("Felladrin/Llama-160M-Chat-v1")
dataset = load_dataset("argilla/openhermes2.5-dpo-binarized-alpha", split="train")
def format(columns): return { "prompt": tokenizer.apply_chat_template(columns["chosen"][:-1], tokenize=False, add_generation_prompt=True), "chosen": f"{columns[chosen][-1][content]}<|im_end|>", "rejected": f"{columns[rejected][-1][content]}<|im_end|>", }
dataset.map(format).select_columns([prompt, chosen, rejected, category, source, chosen_model, rejected_model, rejected_score, chosen_score]).to_parquet("train.parquet")
