activity
20242026
collaborators
Showing 2024Show all

9 papers · 1 filter

cs.CL2024

From Jack of All Trades to Master of One: Specializing LLM-based Autoraters to a Test Set

Mara Finkelstein, Dan Deutsch, Parker Riley +3

As LLMs continue to become more powerful and versatile, human evaluation has quickly become intractable at scale and reliance on automatic metrics has become the norm. Recently, it…

cs.CL2024

Introducing the NewsPaLM MBR and QE Dataset: LLM-Generated High-Quality Parallel Data Outperforms Traditional Web-Crawled Data

Mara Finkelstein, David Vilar, Markus Freitag

Recent research in neural machine translation (NMT) has shown that training on high-quality machine-generated data can outperform training on human-generated data. This work accomp…

cs.CL2024

Mitigating Metric Bias in Minimum Bayes Risk Decoding

Geza Kovacs, Daniel Deutsch, Markus Freitag

While Minimum Bayes Risk (MBR) decoding using metrics such as COMET or MetricX has outperformed traditional decoding methods such as greedy or beam search, it introduces a challeng…

cs.CL2024

LLMRefine: Pinpointing and Refining Large Language Models via Fine-Grained Actionable Feedback

Wenda Xu, Daniel Deutsch, Mara Finkelstein +6

Recent large language models (LLM) are leveraging human feedback to improve their generation quality. However, human feedback is costly to obtain, especially during inference. In t…

cs.CL2024

Learning from others' mistakes: Finetuning machine translation models with span-level error annotations

Lily H. Zhang, Hamid Dadkhahi, Mara Finkelstein +3

Despite growing interest in incorporating feedback to improve language models, most efforts focus only on sequence-level annotations. In this work, we explore the potential of util…

cs.CL2024

Beyond Human-Only: Evaluating Human-Machine Collaboration for Collecting High-Quality Translation Data

Zhongtao Liu, Parker Riley, Daniel Deutsch +4

Collecting high-quality translations is crucial for the development and evaluation of machine translation systems. However, traditional human-only approaches are costly and slow. T…