DENIAHL: In-Context Features Influence LLM Needle-In-A-Haystack Abilities
This work provides a systematic benchmark for analyzing factors influencing recall in long-context models, which is incremental as it expands on previous studies.
The authors tackled the problem of understanding what factors beyond context length affect language models' ability to recall specific information in Needle-in-a-Haystack tasks, finding that features like data type and item size cause performance drops, with GPT-3.5 and LLaMA 2-7B showing stark differences.
The Needle-in-a-haystack (NIAH) test is a general task used to assess language models' (LMs') abilities to recall particular information from long input context. This framework however does not provide a means of analyzing what factors, beyond context length, contribute to LMs' abilities or inabilities to separate and recall needles from their haystacks. To provide a systematic means of assessing what features contribute to LMs' NIAH capabilities, we developed a synthetic benchmark called DENIAHL (Data-oriented Evaluation of NIAH for LLM's). Our work expands on previous NIAH studies by ablating NIAH features beyond typical context length including data type, size, and patterns. We find stark differences between GPT-3.5 and LLaMA 2-7B's performance on DENIAHL, and drops in recall performance when features like item size are increased, and to some degree when data type is changed from numbers to letters. This has implications for increasingly large context models, demonstrating factors beyond item-number impact NIAH capabilities.