Performance, Efficiency and Collapse -- Advantages and Challenges in Offline Post-training of Code LLMs
The findings indicate that, with only a few hours of training, zero-shot code generation performance of LLMs can be substantially improved without online sampling and offline RL produces performance gains across models, although the extent of improvement varies among model families.
Abhinav Anand, Sanjana Reddy Pachika, S. Verma et al.
· 0 citations