[0.1.0] — 2026-05-15

Added

• 26+ PIELM variants with PyTorch-native GPU-aware implementation.
• PINN baselines: VanillaPINN, AdaptivePINN, FourierPINN, MuonPINN.
• Data adapters: CSV, NPZ, PINNacle .dat, PDEBench HDF5.
• PDE operators, collocation samplers, BC/IC constraint blocks.
• Metrics: rmse, mae, relative_l2, max_error, r2_score, MetricsBundle.
• YAML experiment config, CLI (run / sweep / export / list-models).
• ONNX and TorchScript export.
• Sphinx documentation, runnable examples, benchmark scripts.