jiabaoSU
jiabaoSU
Hello, may I ask how this problem was finally solved? I also met the same problem, thank you!
> 你使用的是最新代码版本吗？ > 你使用的是最新代码版本吗？ The program being used is the latest version downloaded from the main branch.
> 你使用的是最新代码版本吗？ This is my program version. Could someone please help me solve the problem?
> 能否提供这张图片，以便我重现问题？ """ Wan2.2 distilled model image-to-video generation example. This example demonstrates how to use LightX2V with Wan2.2 distilled model for I2V generation. """ from lightx2v import LightX2VPipeline import random...
> seq_p_size=2: torchrun --nproc_per_node=2 wan_i2v_distilled.py seq_p_size=4: torchrun --nproc_per_node=4wan_i2v_distilled.py This is how I wrote it. The same issue. Please ask a senior expert to help reproduce it.
> 尝试重新安装 lightx2v 软件包： pip uninstall lightx2v cd /data/su/test/LightX2V pip install -e . I reinstalled lightx2v and ran it with 4 GPUs, but the results were still incorrect. I only...
> Try reinstalling the lightx2v package： pip uninstall lightx2v cd /data/su/test/LightX2V pip install -e . Hello. Could you please help me with this issue?
> If there are cached files from previous versions, delete them first： cd LightX2V rm -rf build/ lightx2v.egg-info/ pip install . Hello, I followed the method you suggested and replaced...
> 首先，拉取最新代码，然后您可以尝试安装我们修改后的 SageAttention 版本（可在https://github.com/ModelTC/SageAttention获取）。或者，将 attn_mode 切换为 flash_attn3。 I have used flash_attn3 and the latest https://github.com/ModelTC/SageAttention program, but the results are still incorrect. I'm seeking your expert advice. https://github.com/user-attachments/assets/ba6b87f7-27c6-4474-a853-ebd53e0875ef