Here's an important one! This release adds Gemma-2, and it also greatly improves model accuracy across the board. It is highly recommended that everyone update to this version immediately to take advantage of these accuracy improvements.
What's Changed
* Fix typo in Main_Demo.ipynb by ianand in https://github.com/TransformerLensOrg/TransformerLens/pull/636
* Add comparing-to-huggingface.ipynb. by joelburget in https://github.com/TransformerLensOrg/TransformerLens/pull/637
* Add tests for gated mlp by anthonyduong9 in https://github.com/TransformerLensOrg/TransformerLens/pull/638
* Match Huggingface MLP implementation exactly. by joelburget in https://github.com/TransformerLensOrg/TransformerLens/pull/641
* Add tests for ActivationCache by FlyingPumba in https://github.com/TransformerLensOrg/TransformerLens/pull/643
* Moved mixtral weights to another module by bryce13950 in https://github.com/TransformerLensOrg/TransformerLens/pull/646
* Fixed weight conversion by bryce13950 in https://github.com/TransformerLensOrg/TransformerLens/pull/648
* Move out pretrained weight conversions by richardkronick in https://github.com/TransformerLensOrg/TransformerLens/pull/647
* Match Huggingface GPT2 implementation *exactly* by joelburget in https://github.com/TransformerLensOrg/TransformerLens/pull/645
* Fix Out bias not being summed in attention component when using 4 bit precision by FlyingPumba in https://github.com/TransformerLensOrg/TransformerLens/pull/654
* Mlp cleanup by bryce13950 in https://github.com/TransformerLensOrg/TransformerLens/pull/652
* Added support for Gemma-2 by neelnanda-io in https://github.com/TransformerLensOrg/TransformerLens/pull/650
* add tests for Attention by anthonyduong9 in https://github.com/TransformerLensOrg/TransformerLens/pull/639