Math17K

New Super-Tuning method enhances LLM fine-tuning efficiency

Researchers have developed a new method called Super-Tuning, which aims to make fine-tuning large language models (LLMs) more efficient. This technique reuses saliency signals from model pruning to identify which parame…