Ask-Anything
Ask-Anything copied to clipboard
OpenGVLab
[CVPR2024 Highlight][VideoChatGPT] ChatGPT with video understanding! And many more supported LMs such as miniGPT4, StableLM, and MOSS.
I noticed that the video dataset such as NextQA has many different instructions for each sample, how can I create this? (in the provided files in `Ask-Anything/video_chat2/DATA.md`)
00 checkpoint = torch.load(model_path, map_location="cpu") 101 else: --> 102 raise RuntimeError("checkpoint url or path is invalid") 103 104 state_dict = checkpoint["model"] RuntimeError: checkpoint url or path is invalid
https://github.com/OpenGVLab/Ask-Anything/blob/332b3ee84c1a5719cf69d26a487ab2f98f85cacb/video_chat2/tasks/retrieval_utils.py#L143 https://github.com/OpenGVLab/Ask-Anything/blob/332b3ee84c1a5719cf69d26a487ab2f98f85cacb/video_chat2/models/videochat2_qformer.py#L372
Hi, Thanks for sharing your code. But may i check with you that FileNotFoundError: [Errno 2] No such file or directory: 'xxxxAsk-Anything/video_chat2/data/vindlu/videos_images/cc3m/a5756e3bb0da5bf21e4824585c12ebfb80e48356faf9cabce3f41b96a410ff9c.jpg' for cc3m image, what is the meaning of...
It seems that the output results of the web demo have become simpler. I want to know what changes have taken place. There were many repeated answers before.
I noticed that sometimes the model would return an empty answer"". However, the empty answer is turned out to be correct by the "check_ans".
video_chat2中的videochat2_it.py代码里的 def encode_img(self, image, instruction)中的instruction的作用具体是什么？ [https://github.com/OpenGVLab/Ask-Anything/blob/main/video_chat2/models/videochat2_it.py#L174](url)
well its not an issue, I just have a question and curious to know Hi, can we use this to generate the youtube shorts videos from the input video ?
Fine-grained Pose (NTU RGBD), Scene Transition (MovieNet), Unexpected action (FunQA), Egocentric Navigation (VLN-CE) Datasets of these tasks don't have QA annotations and it seems that you generate annotations by yourself...
