how to change the script to finetune based on codellama

nlpxucan / WizardLM

LLMs build upon Evol Insturct: WizardLM, WizardCoder, WizardMath

9.06k stars 709 forks source link

how to change the script to finetune based on codellama #236

Open hgcdanniel opened 5 months ago

hgcdanniel commented 5 months ago

T want to change the base model from startcoder to codellama , But I always have this problem, self._tokenizer.enable_truncation(**target) OverflowError: int too big to convert

could you share your script how to finetune on codellama.