Remedy-R: Generative Reasoning for Machine Translation Evaluation without Error Annotations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tan, Shaomu, Mitani, Ryosuke, Choudhary, Ritvik, Wu, Qiyu, Sekiya, Toshiyuki, Monz, Christof
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911331513270272
author Tan, Shaomu
Mitani, Ryosuke
Choudhary, Ritvik
Wu, Qiyu
Sekiya, Toshiyuki
Monz, Christof
author_facet Tan, Shaomu
Mitani, Ryosuke
Choudhary, Ritvik
Wu, Qiyu
Sekiya, Toshiyuki
Monz, Christof
contents Over the years, automatic MT metrics have hillclimbed benchmarks and presented strong and sometimes human-level agreement with human ratings. Yet they remain black-box, offering little insight into their decision-making and often failing under real-world out-of-distribution (OOD) inputs. We introduce Remedy-R, a reasoning-driven generative MT metric trained with reinforcement learning from pairwise translation preferences, without requiring error-span annotations or distillation from closed LLMs. Remedy-R produces step-by-step analyses of accuracy, fluency, and completeness, followed by a final score, enabling more interpretable assessments. With only 60K training pairs across two language pairs, Remedy-R remains competitive with top scalar metrics and GPT-4-based judges on WMT22-24 meta-evaluation, generalizes to other languages, and exhibits strong robustness on OOD stress tests. Moreover, Remedy-R models generate self-reflective feedback that can be reused for translation improvement. Building on this finding, we introduce Remedy-R Agent, a simple evaluate-revise pipeline that leverages Remedy-R's evaluation analysis to refine translations. This agent consistently improves translation quality across diverse models, including Qwen2.5, ALMA-R, GPT-4o-mini, and Gemini-2.0-Flash, suggesting that Remedy-R's reasoning captures translation-relevant information and is practically useful.
format Preprint
id arxiv_https___arxiv_org_abs_2512_18906
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Remedy-R: Generative Reasoning for Machine Translation Evaluation without Error Annotations
Tan, Shaomu
Mitani, Ryosuke
Choudhary, Ritvik
Wu, Qiyu
Sekiya, Toshiyuki
Monz, Christof
Computation and Language
Over the years, automatic MT metrics have hillclimbed benchmarks and presented strong and sometimes human-level agreement with human ratings. Yet they remain black-box, offering little insight into their decision-making and often failing under real-world out-of-distribution (OOD) inputs. We introduce Remedy-R, a reasoning-driven generative MT metric trained with reinforcement learning from pairwise translation preferences, without requiring error-span annotations or distillation from closed LLMs. Remedy-R produces step-by-step analyses of accuracy, fluency, and completeness, followed by a final score, enabling more interpretable assessments. With only 60K training pairs across two language pairs, Remedy-R remains competitive with top scalar metrics and GPT-4-based judges on WMT22-24 meta-evaluation, generalizes to other languages, and exhibits strong robustness on OOD stress tests. Moreover, Remedy-R models generate self-reflective feedback that can be reused for translation improvement. Building on this finding, we introduce Remedy-R Agent, a simple evaluate-revise pipeline that leverages Remedy-R's evaluation analysis to refine translations. This agent consistently improves translation quality across diverse models, including Qwen2.5, ALMA-R, GPT-4o-mini, and Gemini-2.0-Flash, suggesting that Remedy-R's reasoning captures translation-relevant information and is practically useful.
title Remedy-R: Generative Reasoning for Machine Translation Evaluation without Error Annotations
topic Computation and Language
url https://arxiv.org/abs/2512.18906