A Post-Training Enhanced Optimization Approach for Small Language Models

Keke Zhai

This paper delves into the continuous post-training optimization methods for small language models, and proposes a continuous post-training alignment data construction method for s…