6 accepted papers
We investigate how large language models (LLMs) perform latent multi-hop reasoning in prompts like “Wolfgang Amadeus Mozart’s mother’s spouse is”. To analyze this process, we introduce logit flow, an interpretability method that traces how logits propagate across layers and positions toward the fina
Although multimodal large language models (MLLMs) have achieved impressive performance, the multimodal instruction tuning stage often causes catastrophic forgetting of the base LLM’s language ability, even in strong models like Llama3. To address this, we propose Locate-then-Merge, a training-free p
We investigate the mechanism of in-context learning (ICL) on sentence classification tasks with semantically-unrelated labels (“foo”/“bar”). We find intervening in only 1% heads (named “in-context heads”) significantly affects ICL accuracy from 87.6% to 24.4%. To understand this phenomenon, we analy…
We find arithmetic ability resides within a limited number of attention heads, with each head specializing in distinct operations. To delve into the reason, we introduce the Comparative Neuron Analysis (CNA) method, which identifies an internal logic chain consisting of four distinct stages from inp…
Identifying important neurons for final predictions is essential for understanding the mechanisms of large language models. Due to computational constraints, current attribution techniques struggle to operate at neuron level. In this paper, we propose a static method for pinpointing significant neur…
Binary source code matching, especially on function-level, has a critical role in the field of computer security. Given binary code only, finding the corresponding source code improves the accuracy and efficiency in reverse engineering. Given source code only, related binary code retrieval contribut…