Closed ghost closed 5 months ago
Hi @mxjyst ,
You can go with WikiText-2. The calib dataset should be different from the evaluation dataset to demonstrate the generality.
By the way, we updated the codebase recently to support FP4 accuracy evaluation. You can try it with arg "--quant_type fp".
I'll try it later. Thanks a lot.
Hi, I am wandering which dataset should be the calib dataset? I want to evaluate the quantized model on my own dataset. Which dataset to generate the calib data? my own dataset or other public dataset like wikitext?