Minor fix for llama - Githubissues

IST-DASLab / gptq

Code for the ICLR 2023 paper "GPTQ: Accurate Post-training Quantization of Generative Pretrained Transformers".

https://arxiv.org/abs/2210.17323

Apache License 2.0

1.81k stars 145 forks source link

Minor fix for llama #16

Closed Xiuyu-Li closed 1 year ago

Xiuyu-Li commented 1 year ago

It seems that llama.py does not feed position_ids and will throw out an error when running it. This PR addresses this issue and can run the llama script and reproduce results successfully.

efrantar commented 1 year ago

Thanks for fix, I must have run the code with an older HF version.