MetaDelta++: Improve Generalization of Few-shot System Through Multi-Scale Pretrained Models and Improved Training Strategies

News!

• Serve as a baseline adopted for the Cross-Domain MetaDL competition!
• We've removed all the multiprocessing and uncessary parts to make the logic nice and clean. For the full version, you can refer to the branch neurips2021 or aaai2021.

Requirements

Please follow the official website to set-up environments.

Run the code under competition setting

Please follow the official website to run the codes.
For example, cd to the folder of the cd-metadl, and run the following command:

cd path/to/cd-metadl
python -m cdmetadl.run --input_data_dir=public_data --submission_dir=path/to/this/folder --output_dir_ingestion=ingestion_output --output_dir_scoring=scoring_output --verbose=False --overwrite_previous_results=True --test_tasks_per_dataset=10

Remember to replace the path/to/cd-metadl and path/to/this/folder to your settings.

@AutoMLGroup/Meta_Learners