AULAY WANG
AULAY WANG
Hello, I met the similar problem. I use the vailation of webvid-2M, about 5k videos, to finetune the motion lora. For 50 steps, the results are similar like you(different imgs...
I met the same problem. I think it is caused because my CUDA version is 12.1. `fkash-attn=2.5.6` does not work. Does anyone have the solution?
I have solved this problem by [link](https://github.com/kijai/ComfyUI-SUPIR/issues/143) `pip install open-clip-torch==2.24.0`
> I think it is being auto-added during inference without having to manually add to configs. See: > > https://github.com/hpcaitech/Open-Sora/blob/ee909a7d6611bcc9c5cf1ac055a7e9cc74157e09/docs/report_03.md?plain=1#L121 Can you give me an example? Btw, I tried to...
> We use this function https://github.com/hpcaitech/Open-Sora/blob/main/opensora/utils/inference_utils.py#L55 to append camera motion to prompts. > > Just to double check if the `camera_motion` value is actually there, you may want to print...
I met the same problem.
我也有同样的问题，当我在运行finetune或者validation时显示RuntimeError: Unknown model (iformer_small)
The problem still exists, finetune for 1k iters with ~50 videos, lr=[1e-5,1e-3] do not work.
got it! need to fix the `inference.sh`: ``` run_cmd="$environs python sample_video.py --base configs/cogvideox_2b_lora.yaml configs/inference.yaml --seed $RANDOM" ```