Sparse-autoencoder

What's Changed
* Remove tokenizers pin by ArthurConmy in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/116
* Validation and Checkpoint Intervals by efarrell1 in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/117
* Add data generation pipeline tests by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/118
* Fix the linear encoder's initialisation by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/121
* Fix loss reducer type signature by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/120
* Add toml extension to recommended extensions by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/119
* Add tests against Neel's implementation by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/122

New Contributors
* efarrell1 made their first contribution in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/117

**Full Changelog**: https://github.com/ai-safety-foundation/sparse_autoencoder/compare/v0.14.0...v0.15.0

What's Changed
* Add learned activations L0 norm metric by lucyfarnik in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/91
* Refactor pipeline and abstract pipeline to be a single class by bryce13950 in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/113

New Contributors
* lucyfarnik made their first contribution in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/91

**Full Changelog**: https://github.com/ai-safety-foundation/sparse_autoencoder/compare/v0.13.0...v0.14.0

What's Changed
* Make the geometric median dataset optional by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/110
* Improve log names by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/109
* Add replace and zero activations hooks by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/111
* Add a model reconstruction validation metric by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/112


**Full Changelog**: https://github.com/ai-safety-foundation/sparse_autoencoder/compare/v0.12.0...v0.13.0

What's Changed
* Typo in adam_with_reset.py by HoagyC in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/99
* Add resample metrics by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/98
* Fix numerical stability issues with linear encoder by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/100
* Fix maths details in contributing guide by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/101
* Fix store size calc by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/106
* Move other train metrics from pipeline to implement abstract metric c… by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/107
* Replicate Neel's GeLU work by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/108


**Full Changelog**: https://github.com/ai-safety-foundation/sparse_autoencoder/compare/v0.11.0...v0.12.0

What's Changed
* Rename the source model directory by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/94


**Full Changelog**: https://github.com/ai-safety-foundation/sparse_autoencoder/compare/v0.9.0...v0.10.0

What's Changed
* Prevent num_resample_inputs > len(activation_store) by HoagyC in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/69
* no auto formatting in notebooks by jbloomAus in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/70
* Dataset classes by jbloomAus in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/71
* improve demo by jbloomAus in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/72
* Move towards abstract final pattern by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/74
* Docs generation by bryce13950 in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/60
* Docs deploy by bryce13950 in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/81
* removed deploy action by bryce13950 in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/82
* added generation again by bryce13950 in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/83
* Improve optimizer documentation by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/79
* Add abstract model class by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/80
* Cover notebooks with checks by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/84
* Move pipeline to simpler OO approach by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/85
* Make the CD consistent with the poe commands by alan-cooney in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/93

New Contributors
* bryce13950 made their first contribution in https://github.com/ai-safety-foundation/sparse_autoencoder/pull/60

**Full Changelog**: https://github.com/ai-safety-foundation/sparse_autoencoder/compare/v0.8.0...v0.9.0