RecipeNet: A Hierarchical Transformer for Recipe Data
Updated
Updated · arxiv.org · Aug 14
RecipeNet: A Hierarchical Transformer for Recipe Data
1 articles · Updated · arxiv.org · Aug 14
Summary
Researchers have introduced RecipeNet, a hierarchical Transformer model designed to handle complex, variable-schema recipe data.
RecipeNet outperforms existing tabular and transformer-based models on multiple recipe datasets, particularly in predicting procedural steps and masked-step recovery.
By preserving the hierarchical and sequential structure of recipe data, RecipeNet offers improved accuracy and efficiency for applications in manufacturing, materials science, and pharmaceuticals.