jw xiong
jw xiong
It seems to ignore the attention_mask used in DiT forward function. ``` File "/mnt/workspace/Text-to-Video/Open-Sora-Plan/opensora/models/diffusion/diffusion/respace.py", line 130, in __call__ return self.model(x, new_ts, **kwargs) TypeError: forward() missing 1 required positional argument: 'attention_mask'...
可以更新下二维码吗？
虽然我也不太清楚，但你可以发邮件沟通一下
But it seems that these do not prove that the encoder_hidden_states are none and replaced by the hidden_states.
Same question! And many of the downloaded images do not contain human information.
@SSUHan Hi, did you solve the problem?
确实可以训练，但碰到这个问题，发现是 gradient_checkpointing设置为了true导致的，可能是代码库还没有进行修改，来匹配qwen2-vl。 解决方案就是设置gradient_checkpointing=false就行了  
Can you provide the details?