Thank you for your attention to our work.

#1
by jiamingZ - opened

We greatly appreciate your attention to our work and your conversion of the gguf format weights.
This has been a tremendous convenience for the community.

We want to reference these weights on our homepage.
Therefore, we would like to know the details of the conversion and would appreciate it if you could update the README file in this repository.

Thank you for your open source.
I used the open source project links to perform a diffuser conversion on your model based on the WAN category gguf, but I haven't tested it yet

不可能成功,必须用gguf comfyui里面的脚本转换。我的机器内存太低没法转换,要不就转了

不可能成功,必须用gguf comfyui里面的脚本转换。我的机器内存太低没法转换,要不就转了

@ZKong Could you provide the script link? We have a machine and can try to convert it.

https://github.com/city96/ComfyUI-GGUF/tree/main/tools 按照他这个来就行了,很简单

要在comfyui里面用的话,得用kj转换的那个comfyui bf16版本

要在comfyui里面用的话,得用kj转换的那个comfyui bf16版本

@ZKong Do you mean convert the Kj weight in
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/SteadyDancer/Wan21_I2V_SteadyDancer_fp16.safetensors
, instead of ours?

对,你们的应该没法用在comfyui里面,comfyui支持的模型的一些key改了name

这个是diffuser版本,不适配comfyUI原生修改过的模型结构,包括KJ的 运行方式需要用diffuser的from_singlefiles()加载

Sign up or log in to comment