search

zhiyuanhubj / LongRecipe

LongRecipe: Recipe for Efficient Long Context Generalization in Large Language Models
https://arxiv.org/abs/2409.00509
70 stars 4 forks source link

replay_dataset #3

Closed 233function closed 1 month ago

233function commented 2 months ago

你好，第二步训练时的replay_dataset的地址有么

zhiyuanhubj commented 2 months ago

Hi, here are the link for these datasets for replaying.

WizardLM-evol-instruct-70k: https://huggingface.co/datasets/WizardLMTeam/WizardLM_evol_instruct_70k, MagicoderOSS-Instruct-75K: https://huggingface.co/datasets/ise-uiuc/Magicoder-OSS-Instruct-75K