Doubiiu / DynamiCrafter

[ECCV 2024] DynamiCrafter: Animating Open-domain Images with Video Diffusion Priors
Apache License 2.0
2.09k stars 165 forks source link

请问/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin 应该放到本地哪个目录下? #49

Closed g711ab closed 3 months ago

g711ab commented 3 months ago

我的主机不能访问huggingface,/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin 这个文件已经下载到本地,请问应该放到哪个目录下,谢谢!

Doubiiu commented 3 months ago

I am not sure about that, which is related to your own configuration. One possible solution is adding version: "<path to this bin file in your computer>" under the params of cond_stage_config and img_cond_stage_config in each config yaml:

    cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
      params:
        version: "<path to this bin file in your computer>"
        freeze: true
        layer: "penultimate"

    img_cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
      params:
        version: "<path to this bin file in your computer>"
        freeze: true

I am not sure if it works or not and you can have a try.

hben35096 commented 3 months ago

我的主机不能访问huggingface,/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin 这个文件已经下载到本地,请问应该放到哪个目录下,谢谢!

https://www.codewithgpu.com/m/dynamicrafter_mo

hub.zip 解压后放入 /root/.cache/huggingface

如果是Windows系统,估计在: C:\Users\这里是你的用户名.cache\huggingface

g711ab commented 3 months ago

I am not sure about that, which is related to your own configuration. One possible solution is adding version: "<path to this bin file in your computer>" under the params of cond_stage_config and img_cond_stage_config in each config yaml:

    cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
      params:
        version: "<path to this bin file in your computer>"
        freeze: true
        layer: "penultimate"

    img_cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
      params:
        version: "<path to this bin file in your computer>"
        freeze: true

I am not sure if it works or not and you can have a try.

thank you for reply, but it is not work for me

g711ab commented 3 months ago

我的主机不能访问huggingface,/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin 这个文件已经下载到本地,请问应该放到哪个目录下,谢谢!

https://www.codewithgpu.com/m/dynamicrafter_mo

hub.zip 解压后放入 /root/.cache/huggingface

如果是Windows系统,估计在: C:\Users\这里是你的用户名.cache\huggingface

非常感谢,路径已找到 .cache/huggingface/hub/models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/snapshots/de081ac0a0ca8dc9d1533eed1ae884bb8ae1404b/ 但还是不行

Doubiiu commented 3 months ago

我的主机不能访问huggingface,/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin 这个文件已经下载到本地,请问应该放到哪个目录下,谢谢!

https://www.codewithgpu.com/m/dynamicrafter_mo hub.zip 解压后放入 /root/.cache/huggingface 如果是Windows系统,估计在: C:\Users\这里是你的用户名.cache\huggingface

非常感谢,路径已找到 .cache/huggingface/hub/models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/snapshots/de081ac0a0ca8dc9d1533eed1ae884bb8ae1404b/ 但还是不行

Maybe you can try my above solution by changing version path as <your path>/models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/blobs/<the file with filesize of ~4GB>

g711ab commented 3 months ago

.cache/huggingface/hub/models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/

cond_stage_config:
  version: "/home/testaa/.cache/huggingface/hub/models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/blobs/open_clip_pytorch_model.bin"
  target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
  params:
    freeze: true
    layer: "penultimate"

img_cond_stage_config:
  version: "/home/testaa/.cache/huggingface/hub/models--laion--CLIP-ViT-H-14-laion2B-s32B-b79K/blobs/open_clip_pytorch_model.bin"
  target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
  params:
    freeze: true

thank you ,it's my config file,This issue has not been resolved.

Doubiiu commented 3 months ago

First the version should be put under params like what I have mentioned here:

    cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
      params:
        version: "xxxxx"
        freeze: true
        layer: "penultimate"

    img_cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
      params:
        version: "xxxx"
        freeze: true

Second, what files were originally in your folder of blobs (instead of .bin)?

hben35096 commented 3 months ago

@g711ab

image 这些文件都是必要的

g711ab commented 3 months ago

First the version should be put under params like what I have mentioned here:

    cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
      params:
        version: "xxxxx"
        freeze: true
        layer: "penultimate"

    img_cond_stage_config:
      target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
      params:
        version: "xxxx"
        freeze: true

Second, what files were originally in your folder of blobs (instead of .bin)?

thank you very much! it's worked.

https://github.com/Doubiiu/DynamiCrafter/assets/4608210/f6f897c5-6855-4474-853f-852b402dad8f

very smooth

g711ab commented 3 months ago

@g711ab

image 这些文件都是必要的

感谢,已经好了