analiza aiqbench
This study investigates the efficacy of intermediate reasoning traces in large reasoning models (LRMs) specifically for machine translation tasks, moving beyond general capability assessments to analyze structural patterns like planning and verification. The research identifies that reasoning quality is highly dependent on model-specific language preferences and exhibits a non-monotonic relationship with length, suggesting that uniform encouragement of reasoning steps may not always yield optimal translation results. By introducing the Hierarchical Meta-Summarization framework, the authors provide a scalable method to induce granular reasoning structures that adapt to domain variations without relying on predefined taxonomies.
When Does Reasoning Help in Machine Translation? A Hierarchical Analysis of LRM Reasoning Traces
arXiv:2609.21247v1 Announce Type: new Abstract: Large Reasoning Models increasingly use intermediate traces for machine translation, but it remains unclear when such reasoning helps or hurts. We analyze reasoning traces across models, languages, domains, and datasets, focusing on reasoning language, length, and structure. We find that the best reasoning language is model-specific, reasoning length has a non-monotonic relationship with quality, and traces exhibit recurring functional patterns. To uncover these patterns, we introduce Hierarchical Meta-Summarization (HMS), a scalable framework that induces coarse- and fine-grained reasoning structures without predefined taxonomies. HMS reveals a shared organization--understanding/planning, translating/drafting, and refining/verifying--alongside domain-specific variation. Our results suggest that MT reasoning should be controlled in a model-aware, length-aware, and pattern-aware manner rather than uniformly encouraged.
Title:When Does Reasoning Help in Machine Translation? A Hierarchical Analysis of LRM Reasoning Traces
Abstract:Large Reasoning Models increasingly use intermediate traces for machine translation, but it remains unclear when such reasoning helps or hurts. We analyze reasoning traces across models, languages, domains, and datasets, focusing on reasoning language, length, and structure. We find that the best reasoning language is model-specific, reasoning length has a non-monotonic relationship with quality, and traces exhibit recurring functional patterns. To uncover these patterns, we introduce Hierarchical Meta-Summarization (HMS), a scalable framework that induces coarse- and fine-grained reasoning structures without predefined taxonomies. HMS reveals a shared organization--understanding/planning, translating/drafting, and refining/verifying--alongside domain-specific variation. Our results suggest that MT reasoning should be controlled in a model-aware, length-aware, and pattern-aware manner rather than uniformly encouraged.
Computation and Language (cs.CL)
arXiv:2609.21247 [cs.CL]
(or arXiv:2609.21247v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2609.21247 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 221 cuvinte
