Felladrin/ChatML-distilabel-intel-orca-dpo-pairs

---
license: apache-2.0
language:
- en
size_categories:
- 10K<n<100K
---

[argilla/distilabel-intel-orca-dpo-pairs](https://huggingface.co/datasets/argilla/distilabel-intel-orca-dpo-pairs) 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

dataset = load_dataset("argilla/distilabel-intel-orca-dpo-pairs", split="train")

def format(columns):
prompt = f"<|im_start|>user\n{columns['input']}<|im_end|>\n<|im_start|>assistant\n"

if (columns['system']):
prompt = f"<|im_start|>system\n{columns['system']}<|im_end|>\n{prompt}"

return {
"prompt": prompt,
"chosen": f"{columns['chosen']}<|im_end|>",
"rejected": f"{columns['rejected']}<|im_end|>",
}

dataset.map(format).select_columns(['prompt', 'chosen', 'rejected', 'status', 'chosen_score', 'in_gsm8k_train']).to_parquet("train.parquet")
```

许可证：Apache-2.0
语言：
- 英语
样本规模范围：
- 10000 < 样本量 < 100000

本数据集为ChatML格式的[argilla/distilabel-intel-orca-dpo-pairs](https://huggingface.co/datasets/argilla/distilabel-intel-orca-dpo-pairs)，可直接用于[HuggingFace TRL的DPO训练器](https://huggingface.co/docs/trl/main/en/dpo_trainer)。

以下为格式转换所用的Python代码：
python
from datasets import load_dataset

dataset = load_dataset("argilla/distilabel-intel-orca-dpo-pairs", split="train")

def format(columns):
prompt = f"<|im_start|>user
{columns['input']}<|im_end|>
<|im_start|>assistant
"

if (columns['system']):
prompt = f"<|im_start|>system
{columns['system']}<|im_end|>
{prompt}"

return {
"prompt": prompt,
"chosen": f"{columns['chosen']}<|im_end|",
"rejected": f"{columns['rejected']}<|im_end|",
}

dataset.map(format).select_columns(["prompt", "chosen", "rejected", "status", "chosen_score", "in_gsm8k_train"]).to_parquet("train.parquet")

原始信息汇总

数据集概述

基本信息

• 许可证: Apache-2.0
• 语言: 英语
• 数据量: 10K<n<100K

数据格式

• 数据集以ChatML格式提供，适用于HuggingFace TRLs DPO Trainer。

数据处理

• 数据集包含以下字段：

  • input
  • system
  • chosen
  • rejected

• 数据处理代码示例： python from datasets import load_dataset

  dataset = load_dataset("argilla/distilabel-intel-orca-dpo-pairs", split="train")

  def format(columns): prompt = f"<|im_start|>user {columns[input]}<|im_end|> <|im_start|>assistant "

  if (columns[system]):
      prompt = f"<|im_start|>system
  

{columns[system]}<|im_end|> {prompt}"

  return {
      "prompt": prompt,
      "chosen": f"{columns[chosen]}<|im_end|>",
      "rejected": f"{columns[rejected]}<|im_end|>",
  }

dataset.map(format).select_columns([prompt, chosen, rejected, status, chosen_score, in_gsm8k_train]).to_parquet("train.parquet")