Learning from Saturated Data: Signals Beyond Correctness for LLM Training 事件

Learning from Saturated Data: Signals Beyond Correctness for LLM Training arXiv:2606.01436v1 Announce Type: new Abstract: The growing capabilities of large language models (LLMs) have led to the saturation of many benchmarks and training datasets used to improve them. Motivated by this, we investigate whether questions solved with perfect empirical accuracy can nevertheless be used to improve downstream performance. To do so, we replace binary correctness with two sources of more fine-grained q