Shiwei Zhang comments

Results 73 comments of


                                            Shiwei Zhang

problems about i2v demo

Hi, what is the status of your GPU memory usage?

when i run in windows,have a error about NCCL?

Sorry, we haven't tested on the Windows system yet. Currently, we have only developed all the models in a Linux + A100 environment.

when i run in windows,have a error about NCCL?

This issue has been solved by [here](https://github.com/damo-vilab/i2vgen-xl/blob/main/tools/modules/diffusions/diffusion_ddim.py#L15). You can clone the code again.

Exception: Failed to invoke function <function inference_i2vgen_entrance at 0x7f56eb98d5e0>, with indices should be either on cpu or on the same device as the indexed tensor (cpu)

We found that this issue may arise in different environments. It has been fixed, you can pull the latest code again.

Exception: Failed to invoke function <function inference_i2vgen_entrance at 0x7f56eb98d5e0>, with indices should be either on cpu or on the same device as the indexed tensor (cpu)

Or you can change the code, refer to [here](https://github.com/damo-vilab/i2vgen-xl/blob/main/tools/modules/diffusions/diffusion_ddim.py#L15)

Any hope for Intel Arc/IPEX Support?

Thank you for your interest in our work, but we may not be able to support it at this time.

snapshot_download failed to finish the download

Sorry, you can also try downloading the model from huggingface. https://huggingface.co/damo-vilab/i2vgen-xl/tree/main

Running inference without CPU and GPU on machine

Thank you for your interest in our work. Currently, we do not support CPU inference because it would be very slow, and some components also rely on the GPU.

about DreamTalk code？

Thank you for your interest in DreamTalk. We are also in preparation. Considering that the code design is different, DreamTalk will create a new repository separately. When it is ready...

ERROR: No matching distribution found for motion-vector-extractor==1.0.6

Hello, for the current version, you do not need to install the motion vector. However, the upcoming open-source models will require support for motion vectors.