diff --git a/muNet/NAX.ipynb b/muNet/TransformerExpansions.ipynb similarity index 98% rename from muNet/NAX.ipynb rename to muNet/TransformerExpansions.ipynb index 1692f2007ff9..2193cc4def09 100644 --- a/muNet/NAX.ipynb +++ b/muNet/TransformerExpansions.ipynb @@ -22,9 +22,9 @@ "id": "7BQOCQE5tZiU" }, "source": [ - "# Neural Architecture EXpansion (NAX): Incremental scaling of Transformer-based Neural Networks\n", + "# Composable Function-preserving Expansions for Transformer Architectures\n", "\n", - "This notebook contains implementations of the six function-preserving transformations of transformer-based models proposed in \"Neural Architecture EXpansion (NAX): Incremental scaling of Transformer-based Neural Networks\". We provide a basic implementation of a generic transformer architecture and show that each transformation is function-preserving, both for individual architectural components and the whole transformer model, as well as for individual transformations and combinations of transformations.\n" + "This notebook contains implementations of the six function-preserving transformations of transformer-based models proposed in \"Composable Function-preserving Expansions for Transformer Architectures\". We provide a basic implementation of a generic transformer architecture and show that each transformation is function-preserving, both for individual architectural components and the whole transformer model, as well as for individual transformations and combinations of transformations.\n" ] }, {