question about training

[CrazyBrick]

Can you share the trained model weights or provide non distributed training settings (single machine single GPU)
How long does it take for you to train once？
Thank you very much

[CrazyBrick]

FAVDBench/AVLFormer/src/datasets/vl_dataloader.py", line 21, in build_dataset
assert op.isfile(yaml_file), f"{yaml_file} does not exists"
AssertionError: datasets/train_32frames.yaml does not exists

"train_yaml": "train_32frames.yaml",
"val_yaml": "val_32frames.yaml",

These two files are missing. Do you have time to make up for the missing files.Thanks in advance.

[CrazyBrick]

Hi, I generated files such as train_img. tsv using the path, and after using run. sh and installing ffmpeg, I successfully generated files such as frame_tsv/* _32frames. img. tsv. I miss your train_ The 32frames.yaml file should probably be

Caption: 'train. caption. tsv',
Caption_ Coco_ Format: 'train. capture'_ Coco_ Format. json ',
Caption_ Linelist: 'train. caption. linelist. tsv',
Img:'frame_ Tsv/train_ 32frames. img. tsv ',
Label: 'train. label. tsv'
Audio: 'train'_ Mp3. hdf '"

I don't know how to obtain these contents, except for img.
I do hope you can help with it. It would be even better if you could open the pretrained model as soon as possible

[AdrienneDeganutti]

@CrazyBrick Hello, I'm currently trying to implement the FAVD model and have just encountered this same issue. Did you ever manage to successfully generate these two missing yaml files?

[CrazyBrick]

@CrazyBrick Hello, I'm currently trying to implement the FAVD model and have just encountered this same issue. Did you ever manage to successfully generate these two missing yaml files?

Sorry, I didn't try successfully. And I sent an email to the author, but he didn't modify the code or answer questions afterwards.

[swapb94]

@CrazyBrick @AdrienneDeganutti
Has anyone managed to get the yaml files or any possible workaround??

[XuyangShen]

Hi, I generated files such as train_img. tsv using the path, and after using run. sh and installing ffmpeg, I successfully generated files such as frame_tsv/* _32frames. img. tsv. I miss your train_ The 32frames.yaml file should probably be

Caption: 'train. caption. tsv',
Caption_ Coco_ Format: 'train. capture'_ Coco_ Format. json ',
Caption_ Linelist: 'train. caption. linelist. tsv',
Img:'frame_ Tsv/train_ 32frames. img. tsv ',
Label: 'train. label. tsv'
Audio: 'train'_ Mp3. hdf '"

I don't know how to obtain these contents, except for img. I do hope you can help with it. It would be even better if you could open the pretrained model as soon as possible

Sorry for the late reply. We uploaded the processed files and detailed instructions in README.md.

Again, sorry for the late. Feel free to comment on any issues!

[XuyangShen]

@CrazyBrick @AdrienneDeganutti Has anyone managed to get the yaml files or any possible workaround??

Sorry for the late reply. We uploaded the processed files and detailed instructions in README.md.

Again, sorry for the late. Feel free to comment on any issues!

[XuyangShen]

@CrazyBrick Hello, I'm currently trying to implement the FAVD model and have just encountered this same issue. Did you ever manage to successfully generate these two missing yaml files?

Sorry for the late reply. We uploaded the processed files and detailed instructions in README.md.

Again, sorry for the late. Feel free to comment on any issues!