BERNN-MSMS
Minimal README for quick usage.
Longer historical content is kept in LEGACY_README.md.
Install
pip install bernn
Basic usage
from bernn import TrainAEClassifierHoldout
trainer_cls = TrainAEClassifierHoldout
trainer = trainer_cls(config=bernn_config, log_metrics=True, keep_models=False)
# Train and predict in one call
preds_encoded = trainer.fit_predict(
X_train,
y_train,
X_test=X_test,
y_test=y_test,
groups_train=batches_train,
groups_test=batches_test,
cross_validation=False,
cross_test=False,
)
# Decode predictions back to original labels
preds = trainer.predict(X_test)
Important runtime contract:
- groups_train is mandatory.
- If X_test is provided, groups_test is mandatory.
Important parameters
Focus on these first:
- optimize_hyperparams: enable/disable Ax optimization.
- n_trials: number of optimization trials.
- fixed_hyperparams: force values and remove them from search.
- n_repeats: number of holdout repeats.
- n_layers, layer1: classifier depth and width seed.
- dloss: domain loss mode.
- warmup, n_epochs: core training schedule.
- device: cpu/cuda target.
- scaler, bs: preprocessing and batch size.
Official documentation
- Full reference: OFFICIAL_DOCUMENTATION.md
- Full parameter catalog: TRAINING_PARAMETERS.md
- Minimal runnable examples notebook (4 variants): tutorials/minimal_examples.ipynb
- Optimized all-config notebook (TrainAEClassifierHoldout): tutorials/optimized_classifier_holdout_all_configs.ipynb
- Optimized all-config notebook (TrainAEThenClassifierHoldout): tutorials/optimized_ae_then_classifier_holdout_all_configs.ipynb
- Historical CLI-heavy guide: LEGACY_README.md
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
bernn-1.0.2.tar.gz
(298.6 kB
view details)
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
bernn-1.0.2-py3-none-any.whl
(319.1 kB
view details)
File details
Details for the file bernn-1.0.2.tar.gz.
File metadata
- Download URL: bernn-1.0.2.tar.gz
- Upload date:
- Size: 298.6 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.11.15
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
7389697f1f04af3053001b6a9580aa8e702206d48af42c3798814d1f2ca68b98
|
|
| MD5 |
7f18f67dbddd8db46d86bc06d530797c
|
|
| BLAKE2b-256 |
efe6e9eb0c7db7f1f12216cadabf907ea9875b679d291c03734c46711fcdf27a
|
File details
Details for the file bernn-1.0.2-py3-none-any.whl.
File metadata
- Download URL: bernn-1.0.2-py3-none-any.whl
- Upload date:
- Size: 319.1 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.11.15
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
f45029590fc371d649399c90e276c6ce6c92e254310f38ba55b99b15041007b1
|
|
| MD5 |
bbcdde5f672cc5c0a03cb64b3e39c528
|
|
| BLAKE2b-256 |
70b02d5d9163bfe6bb08524485c217f04dfd63ab78894ac492b0252d354b7515
|
