Di Politecnico, Paolo Garza, Daniele Rege Cambrin, Torino, Mattia Ottoborgo
We lifted 1 functions out of this paper's own repositories and ran 0 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| DarthReca/losses-cook | canonical | 0 of 1 |
| Function | Status | Where it lives |
|---|---|---|
| preprocess_logits_for_metrics | Not yet run | DarthReca/losses-cook/improved_loss.py code served (permissive licence) · get_code("db0916f0450c26db") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
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 language generation metrics and recipe-specific metrics, we find that our loss significantly improves ingredient-and actionlevel 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.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2601.02531")
get_code_for_paper("2601.02531")
have("2601.02531")
Connect an agent — have() is free.