Dataset · Magpie-Align
Magpie Chinese instructions for Qwen2
Magpie-Align/Magpie-Qwen2-Pro-200K-Chinese
200k Chinese instruction-response pairs for aligning and training language models.
What's inside
The dataset contains Chinese instruction-response pairs for training and aligning large language models. The examples were collected automatically using the Magpie method from the open Qwen2 model, without manual annotation. The title indicates about 200k examples; the data is presented in text and tabular formats.
What you can do with it
- Supervised fine-tuning (SFT) of a Chinese language model
- Evaluating instruction-following in Qwen2
- Comparing with other instruction datasets
- Tasks
- Question answering
- Languages
- Chinese
- Size
- 100K–1M rows
- Downloads this month
- 1.4K
- Likes
- 84
Models trained on it
- Magpie-Align/Llama-3-8B-Magpie-Align-SFT-v0.3Text generation 9.3K
- Magpie-Align/Llama-3-8B-Magpie-Align-v0.3Text generation 9.2K
- mradermacher/Mistral-Nemo-NT-Ko-12B-sft-i1-GGUF 1.8K
- mradermacher/Llama-3-8B-Magpie-Mix-SFT-v0.1-i1-GGUF 1.5K
- mradermacher/Llama-3.1-8B-Reason-Blend-888k-i1-GGUF 1.3K
- mradermacher/Llama-3-8B-Magpie-Mix-SFT-v0.1-GGUF 495
- mradermacher/Mistral-Nemo-NT-Ko-12B-sft-GGUF 371
- mradermacher/Llama-3.1-8B-Reason-Blend-888k-GGUF 228
How to load it
from datasets import load_dataset
ds = load_dataset("Magpie-Align/Magpie-Qwen2-Pro-200K-Chinese")