Papers by Daniele Rege Cambrin
Losses that Cook: Topological Optimal Transport for Structured Recipe Generation (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing work on cooking recipes relies on cross-entropy, but it does not address holistic composition of ingredient sets and numerical aspects of recipes. |
| Approach: | They propose a topological loss that represents ingredient lists as point clouds in embedding space . they show that the Dice loss excels in time/temperature precision . |
| Outcome: | The proposed model improves ingredient- and action-level metrics while preserving time/temperature precision. |
Beyond Accuracy Optimization: Computer Vision Losses for Large Language Model Fine-Tuning (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated impressive performance across various tasks, but current training approaches combine cross-entropy loss with extensive data, human feedback, or ad hoc methods to enhance performance. |
| Approach: | They evaluate the use of established semantic segmentation loss functions in natural language generation to create a versatile, practical, and scalable solution for fine-tuning different architectures. |
| Outcome: | The proposed models improve on Math Word Problems and question answering tasks without additional data or human feedback. |