Skip to content

v1.5.0 - Multiclass training, callbacks, and tooling

Choose a tag to compare

View all tags
@SavioCodes SavioCodes released this 21 Mar 21:59
· 20 commits to main since this release
v1.5.0
41ccadb

Highlights

  • Added multiclass classification with softmax and categorical_crossentropy.
  • Added training regularization with L2, dropout, and gradient clipping.
  • Added reusable callbacks: EarlyStopping, History, CSVLogger, and ModelCheckpoint.
  • Added ModelConfig and TrainingConfig for a cleaner, more organized training API.
  • Added new utilities for multiclass datasets, one-hot encoding, richer metrics, and better plots.
  • Added scripts/benchmark.py and a new multiclass example.
  • Improved project quality with mypy, coverage, pre-commit, stronger CI, and expanded tests.
  • Rewrote the README and docs to better teach the codebase and workflow.

Validation

  • python -m ruff check src tests scripts examples
  • python -m mypy src
  • python -m pytest -q
  • python scripts/evaluate.py --seed 42 --epochs 300 --samples 240
  • python scripts/benchmark.py --mode binario --samples 180 --epochs 80
  • python scripts/benchmark.py --mode multiclasse --samples 180 --epochs 80
Assets 2
Loading