Long-context Language Models Are Not Good At ALL Retrieval Tasks Without Sufficient Steps (2410.04422v8)

Abstract: Long-context LLMs (LCLMs), characterized by their extensive context window, are becoming popular. However, despite they are nearly perfect at standard long-context retrieval tasks, our evaluations demonstrate they are not good at 2 basic cases, "multi-matching retrieval," and "logic-based retrieval", which are beyond LCLMs' ability boundary. But we find they can be well addressed with a sufficient number of reasoning steps, guided by specific CoT prompts, indicating the necessity of combining long-context tasks with CoT methods for more advanced long context handling. However, current CoT methods are too time-consuming, when the context is very long, which means efficient long-context handling still has a long way to go.