ppl.pmx

ppl.pmx copied to clipboard

`\{softmax}` is not supported in KaTeX, a change to `softmax` can fix it. 

syheliel

Good first issue

3

Hi guys, I see your project in OSPP. I'm wondering are there any good issues for kick start?

syheliel

文档内显示支持qwen1.5，但实际转换报错。

1

使用的model_zoo/qwen/huggingface/ConvertWeightToOpmx.py，py文件内param参数全都是对应qwen而非qwen1.5设置的。无法找到qwen1.5权重转换对应的脚本。

poemon1

add llama pipeline parallel, do not need to do extra split model. It split tp and pp while loading model

Vincent-syr

support int4 weight only quant for llama3

4

Support int4 weight only quantization for Llama3 1. Define the weight only layer in ModelParallel.py 2. Define ConvertWeightToOpmx.py and add quant here 3. Update Dynamic Static modeling for quant model...

yinfan98

为什么使用llama脚本转换qwen1.5权重文件，转出来的文件是model.pth和opmx_params.json两个文件

6

Flynn-Zh

Jzz24

[Eval] feat: add ppl eval and setup.py for ppl.pmx

2

add setup.py for opmx

yinfan98