← Search

Zijin Gu

2 accepted papers

2026

Closing the Gap Between Text and Speech Understanding in LLMs

ICLR 2026poster

Large Language Models (LLMs) can be adapted to extend their text capabilities to speech inputs. However, these speech-adapted LLMs consistently underperform their text-based counterparts—and even cascaded pipelines—on language understanding tasks. We term this shortfall the text–speech understanding…

Cited by 0SourcecodeScholar
2025

Theory, Analysis, and Best Practices for Sigmoid Self-Attention

ICLR 2025poster

Attention is a key part of the transformer architecture. It is a sequence-to-sequence mapping that transforms each sequence element into a weighted sum of values. The weights are typically obtained as the softmax of dot products between keys and queries. Recent work has explored alternatives to soft…