Tencent/HunyuanDiT

trt部署时,模型导出onnx失败

Opened this issue · 2 comments

Describe the bug
A clear and concise description of what the bug is.
变量style如何处理
企业微信截图_17224141856142
image

Reproduction
按照教程执行,使用hunyuandit提供docker, 硬件A800,driver version 525.85.12
wget https://dit.hunyuan.tencent.com/download/HunyuanDiT/hunyuan_dit_cu11.tar
https://hf-mirror.com/Tencent-Hunyuan/TensorRT-libs/blob/main/README_zh.md

Tasks

Preview Give feedback
No tasks being tracked yet.

@linuxmi Which model did you use then? I met the same issue when using v1.2(needs no style or style_cond) and changed model to v1.1 and another issue came up LOL.

遇到了相同的问题,当用v1.2的时候修改export_onnx.py 将ims和style改成None,遇到了另一个报错
image