Get the latest tech news
The Role of Feature Normalization in Ijepa
Training IJEPA image encoders for the masses. Contribute to theAdamColton/elucidating-featurenorm-ijepa development by creating an account on GitHub.
Download the training datasets and NYU-Depth tar files: uv run download_dataset.py This consumes ~22GB of VRAM and takes 116 hours (assuming validation logging is turned off): uv run main.py --config conf/small.yaml Or plot the losses of a pretrained model: uv run main.py --config /path/to/checkpoint/config.yaml --conf.resume_checkpoint_path /path/to/checkpoint/checkpointfile.pt --conf.mode plot-sample-losses
Or read this on Hacker News