QuantStack

/

Wan2.1_14B_VACE-GGUF

like 242

Follow

QuantStack 1.84k

Thanks for great working!!

I want to load model gguf to VACE(https://github.com/ali-vilab/VACE), fist of all, I check the layer, that all right.
Howevers, when I load weight to VACE I get error in 14B Q8_0.gguf

Traceback (most recent call last):
  File "/home/tuphamkhactu1221/VACE/vace/vace_wan_inference.py", line 367, in <module>
    main(args)
  File "/home/tuphamkhactu1221/VACE/vace/vace_wan_inference.py", line 274, in main
    wan_vace = WanVace(
  File "/home/tuphamkhactu1221/VACE/vace/models/wan/wan_vace.py", line 185, in __init__
    set_module_tensor_to_device(
  File "/opt/conda/envs/venv/lib/python3.10/site-packages/accelerate/utils/modeling.py", line 298, in set_module_tensor_to_device
    raise ValueError(
ValueError: Trying to set a tensor of shape torch.Size([5120, 5440]) in "weight" (which has shape torch.Size([5120, 5120])), this looks incorrect.

My question: Which model did you use to convert to gguf?
If it is vace wan 2.1, I would really appreciate your help to check why it has error likes me.
Thank you very much