revert the gptq format code to fix the regression

intel / auto-round

Advanced Quantization Algorithm for LLMs. This is official implementation of "Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs"

https://arxiv.org/abs/2309.05516

Apache License 2.0

132 stars 18 forks source link

revert the gptq format code to fix the regression #168

Closed wenhuach21 closed 2 weeks ago

wenhuach21 commented 2 weeks ago

The GPTQ API parses the quantize_config.json while Transformers parses the config.json