Datasets:

zake7749
/

kyara-chinese-math-sft-s0-30K

Tasks:

Question Answering

Text Generation

Modalities:

Formats:

Languages:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Kyara: Knowledge Yielding Adaptive Retrieval Augmentation for LLM Fine-tuning

🤗 Hugging Face ｜ 🚀Github ｜ 📑 Paper ｜ 📖 English ｜ 📖 Chinese

kyara

This is a dataset for training Kyara.

We applied PersonaHub to generate 50K math problems and used Gemini-1.5-Flash to filter out data with obvious errors in calculation and reasoning. However, this dataset still contains some computational and logical errors; please use it with caution.

Downloads last month: 9

Models trained or fine-tuned on zake7749/kyara-chinese-math-sft-s0-30K

zake7749/gemma-2-2b-it-chinese-kyara-dpo

Text Generation • 3B • Updated Oct 28, 2024 • 44 • 15