Dataset Card for "Fusang-V1"
"A large curation of instruction-tuning datasets for better bilingual and long-range LLMs"
Fusang-V1 is a diverse and large Instruction-Tuning dataset like teknium/OpenHermes-2.5. On top of teknium/OpenHermes-2.5, Fusang-V1is focused on more versatile ability and Chinese support.
Features
Covering wide range of tasks like math, code, roleplay, function calling, etc with over 1.2M base and 140K long samples, all in same format.… See the full description on the dataset page: https://huggingface.co/datasets/wenbopan/Fusang-v1.
