Research and Deployment at the Cutting Edge of AI

LILT's Applied AI team does basic research at the cutting edge of data, models, and deployment. With LILT, customers are future-proofed as AI develops.

Large Language Models

LILT's V3 model series are the world's first and only models that continuously learn business context from human feedback. They can be deployed securely on-prem and in fully air-gapped environments.

Related Research

RESEARCH

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

UHLIG, WUEBKER, REINAUER, DENERO (2024)

Research and Deployment at the Cutting Edge of AI

Large Language Models

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

Neural Machine Translation Models Can Learn to be Few-shot Learners

Measuring Immediate Adaption Performance for Neural Machine Translation

Compact personalized models for neural machine translation

Hierarchical Incremental Adaptation for Statistical Machine Translation

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

Neural Machine Translation Models Can Learn to be Few-shot Learners

Measuring Immediate Adaption Performance for Neural Machine Translation

Compact personalized models for neural machine translation

Hierarchical Incremental Adaptation for Statistical Machine Translation

Post-Training

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

Automatic Correction of Human Translations

Automatic Bilingual Markup Transfer

End-to-End Neural Word Alignment Outperforms GIZA++

Adding Interpretable Attention to Neural Translation Models Improves World Alignment

Models and Inference for Prefix-Constrained Machine Translation

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

Automatic Correction of Human Translations

Automatic Bilingual Markup Transfer

End-to-End Neural Word Alignment Outperforms GIZA++

Adding Interpretable Attention to Neural Translation Models Improves World Alignment

Models and Inference for Prefix-Constrained Machine Translation

Mixed-Initiative Agentic Systems

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

The Impact of Text Presentation on Translator Performance

Translation technology research and human-computer interaction

Predictive Translation Memory: a Mixed-Initiative System for Human Language Translation

Human Effort and Machine Learnability in Computer Aided Translation

The Efficacy of Human Post-Editing for Language Translation

Cross-lingual Human-Preference Alignment for Neural Machine Translation with Direct Quality Optimization

The Impact of Text Presentation on Translator Performance

Translation technology research and human-computer interaction

Predictive Translation Memory: a Mixed-Initiative System for Human Language Translation

Human Effort and Machine Learnability in Computer Aided Translation

The Efficacy of Human Post-Editing for Language Translation

Start your AI journey with LILT today