Language surgery in multilingual Large Language Models
A novel method that leverages latent injection to enable cross-lingual language control and mitigate language confusion.
Large Language Models (LLMs) have demonstrated remarkable generalization capabilities across tasks and languages, revolutionizing natural language processing. This paper investigates the naturally emerging representation alignment in LLMs, particularly in the middle layers, and its implications for disentangling language-specific and language-agnostic information. We empirically confirm the existence of this alignment, analyze its behavior in comparison to explicitly designed alignment models, and demonstrate its potential for language-specific manipulation without semantic degradation. Building on these findings, we propose Inference-Time Language Control (ITLC), a novel method that leverages latent injection to enable precise cross-lingual language control and mitigate language confusion in LLMs. Our experiments highlight ITLC’s strong cross-lingual control capabilities while preserving semantic integrity in target languages. Furthermore, we demonstrate its effectiveness in alleviating the cross-lingual language confusion problem, which persists even in current large-scale LLMs, leading to inconsistent language generation. This work advances our understanding of representation alignment in LLMs and introduces a practical solution for enhancing their cross-lingual performance.
Latest publications
Confidence-based response abstinence: LLM trustworthiness
A method for confidence estimation in RAG systems that aligns closely with the correctness of LLM outputs.
EMNLPGRAID: Synthetic data generation with geometric constraints and multi-agentic reflection for harmful content detection
A novel pipeline that leverages Large Language Models (LLMs) for dataset augmentation.
EMNLPA comparison of strategies for RAG
Evaluation and comparison of multiple RAG fine-tuning strategies.
EMNLP