Losses that Cook: Topological Optimal Transport for Structured Recipe Generation

TL;DR

This paper introduces a topological loss for recipe generation that improves ingredient accuracy and procedural coherence by representing ingredients as point clouds in embedding space.

Contribution

It proposes a novel topological loss function for structured recipe generation, enhancing ingredient and procedural accuracy over standard methods.

Findings

The topological loss improves ingredient and action-level metrics.

Dice loss enhances time and temperature precision.

Mixed loss achieves balanced trade-offs with human preference support.

Abstract

Cooking recipes are complex procedures that require not only a fluent and factual text, but also accurate timing, temperature, and procedural coherence, as well as the correct composition of ingredients. Standard training procedures are primarily based on cross-entropy and focus solely on fluency. Building on RECIPE-NLG, we investigate the use of several composite objectives and present a new topological loss that represents ingredient lists as point clouds in embedding space, minimizing the divergence between predicted and gold ingredients. Using both standard NLG metrics and recipe-specific metrics, we find that our loss significantly improves ingredient- and action-level metrics. Meanwhile, the Dice loss excels in time/temperature precision, and the mixed loss yields competitive trade-offs with synergistic gains in quantity and time. A human preference analysis supports our finding, showing our model is preferred in 62% of the cases.