[x] Added possibility to download model weights for any Malpolon model given a URL and a few file paths
Added model weight download info for the MultiModelEnsemble (MME) model. The example experiment file of MME now automatically downloads the weights from Seafile if not already on disk, via a new download_weights key in the config file
[x] Updated the way checkpoint_path is passed on to models. Added an attribute checkpoint_path for all Malpolon models
Updated every examples consequently
[x] Added Malpolon as (local) model provider
[x] Created new module malpolon.models.custom_models which will host custom models proposed by Malpolon
Split classes from geolifeclef2024_multimodal_ensemble.py to glc2024_multimodal_ensemble_model.py and glc2024_pre_extracted_prediction_system.py in custom_models to prevent circular import from malpolon.models.model_builder after adding Malpolon as (local) provider
Others
GLC22 examples in benchmark and custom_train have been updated to include an inference run option. This led to changing the return values of the class getter for the test dataset. The class now always return a {data, label} pair, with label of value -1 for test dataset (inference run)
:memo: Changelog
Main changes
download_weights
key in the config filecheckpoint_path
is passed on to models. Added an attributecheckpoint_path
for all Malpolon modelsmalpolon.models.custom_models
which will host custom models proposed by Malpolongeolifeclef2024_multimodal_ensemble.py
toglc2024_multimodal_ensemble_model.py
andglc2024_pre_extracted_prediction_system.py
incustom_models
to prevent circular import frommalpolon.models.model_builder
after adding Malpolon as (local) providerOthers
benchmark
andcustom_train
have been updated to include an inference run option. This led to changing the return values of the class getter for thetest
dataset. The class now always return a{data, label}
pair, withlabel
of value-1
fortest
dataset (inference run)malpolon/tests/test_examples.py
accordingly:white_check_mark: Checklist