Czm369

Results 21 comments of Czm369

make browse_dataset.py faster~

Maybe you can set `--not-show`.

make browse_dataset.py faster~

Some parallel or multilinear speedups are not currently supported in `browse_dataset.py`

docker mmvc hard coded pytorch and cuda version

Thanks for your suggestions.

Reproduce results (DINO) on 10% label data on coco dataset

https://huggingface.co/czm369/MixPL/tree/main/mixpl_dino-4scale_r50_90k_coco-s1-p10.py

TOPK query selection out of range when training MixPL over DINO

It seems that `img_shape` is too small.

TOPK query selection out of range when training MixPL over DINO

Sorry, it is the debug config. It should be `scale = [(1333, 400), (1333, 1200)]` .

The result I reproduced is almost 0.

More details, such as config and log, please.

The result I reproduced is almost 0.

`val_interval=5000`

where is the code of labeled resampling

https://github.com/Czm369/MixPL/blob/main/tools/misc/split_coco.py

where is the code of labeled resampling

You can refer to https://github.com/Czm369/MixPL/blob/410f044972a427b67f34393fc8c9314e04add8a2/configs/_base_/datasets/lvis_v0.5_instance.py#L48