Lost in the Passage: Passage-level In-context Learning Does Not Necessarily Need a "Passage"

원문 초록

By simply incorporating demonstrations into the context, in-context learning (ICL) enables large language models (LLMs) to yield awesome performance on many tasks. In this paper, we focus on passage-level long-context ICL for generation tasks and find that LLMs cannot learn the intrinsic relationships between the demonstration passage and the generation output. We conduct experiments with different LLMs on two typical generation tasks including single-document QA and distractor generation, demonstrating that even a completely meaningless demonstration passage with 1/4 length achieves much better performance than the original full passage. Analysis via attention score reveals that LLMs pay little attention to passages compared to other components in prompt and little attention flows from the passage to other parts of the demonstration, which further confirms our finding. Additionally, experiments on context compression indicate that compression approaches proven effective on other long-context tasks are not suitable for passage-level ICL, since simply using shorter meaningless demonstration passages has achieved competitive performance.