METEOR: Evolutionary Journey of Large Language Models from Guidance to Self-Growth
This addresses the problem of inefficient model evolution for AI practitioners, though it appears incremental as it builds on existing feedback-based learning concepts.
The paper tackles the lack of a unified method for guiding the evolution of large language models from no domain knowledge to expertise, proposing the METEOR method with three training phases that significantly improve accuracy, completeness, relevance, coherence, and reliability in domain-specific tasks.
Model evolution enables learning from feedback to refine experiences and update skills, transforming models from having no domain knowledge to becoming domain experts. However, there is currently no unified and effective method for guiding this evolutionary process. To address this gap, we propose the Meteor method, which includes three training phases: weak-to-strong data distillation, iterative training, and self-evolution strategies. Each phase maximizes the model's inherent domain capabilities, allowing it to autonomously refine its domain knowledge and enhance performance. Experiments demonstrate that our approach significantly improves accuracy, completeness, relevance, coherence, and reliability across domain-specific tasks.