ihelix
Copyright 2026 Nathan

Licensed under the Apache License, Version 2.0 (the "License"); you may not use
these files except in compliance with the License. A copy of the License is in
the LICENSE file.

--------------------------------------------------------------------------------

This product includes software developed at Hugging Face, Inc.

`ihelix/kernels.py` carries over the RMS normalization layers, the depthwise
causal convolution, the chunkwise-parallel and single-step gated delta-rule
routines, the landmark pooler and `repeat_kv` from the `helix-lm` package, which
in turn follows the reference implementations in HuggingFace Transformers
(https://github.com/huggingface/transformers), licensed under the Apache
License, Version 2.0.

Original to this project: the geometry layer (metric-space embedding of
spherical, periodic, linear and logarithmic axes, and the local frames built on
them), the grid layer (neighbour graphs and principal-axis spatial hierarchies),
geodesic neighbourhood attention with compactly supported physical receptive
fields and quadrature weighting, the content-addressed regional index, the
cross-grid reader, and the encode-process-decode model built from them.
