StreamingT2V

StreamingT2V copied to clipboard

[It's a bug] It's not working on huggingface

1

I tried this on huggingface.co space to test but it doesn't work and nothing worked on my mobile phone. You can try here: https://huggingface.co/spaces/PAIR/StreamingT2V 

Mistermango23

Please save Github some bandwidth and list average VRAM usage/minimum VRAM requirements per generation type on the home page. Thanks!

MackNcD

There is a bug about the image SVD inference file

7

 

zhanghongyong123456

manyuan97

Hi, is there any config/settings to reproduce the videos shown here as demos? My brief tests with the default settings in `inference.py` yield bad results, see: https://github.com/Picsart-AI-Research/StreamingT2V/assets/6610675/3f9fb6da-5fd3-4509-a48c-f526ec1cf9c8

Ericxgao

File: [model_init.py](https://github.com/Picsart-AI-Research/StreamingT2V/blob/main/t2v_enhanced/model_init.py) pipe.enable_model_cpu_offload() return pipe.to(device) It seems after enabling CPU offloading option, model is send to CUDA device. It is done so in a number of model initializations. It seems...

ajikmr

About GPU configuration

1

Very good job, what kind of GPU configuration do you need? What is the training time?

NuanBaobao

Thank you for your great contributions! I noticed in your paper that the model is trained with a dataset from publicly available sources. Could you possibly provide further details about...

hyc9

MaxRetryError - modelscope doesnt download

2

MaxRetryError: None: Max retries exceeded with url: http://www.modelscope.cn/api/v1/models/damo/ Video-to-Video/repo?Revision=v1.1.0&FilePath=non_ema_0035000.pth (Caused by ConnectionError(ReadTimeoutError("HTTPC onnectionPool(host='www.modelscope.cn', port=80): Read timed out."))) Please help with this error. Also, I couldnt see Animatediff and SVD even...

snehasree-sony

VideoCrafter2 is a strong T2V model. Did you apply StreamingT2V to it?

Yuanshi9815