v0.0.2
- added
eval_tasks
module to support evaluate model's performance on predefined down-stream tasks before and after quantization - fixed some bugs when using
LLaMa
model - fixed some bugs when using models that required
position_ids
eval_tasks
module to support evaluate model's performance on predefined down-stream tasks before and after quantizationLLaMa
modelposition_ids