Yuzhe Gu

I am an ML researcher at UC Santa Cruz, advised by Prof. Yi Zhang.

My current research focuses on robotics, with a particular interest in developing agentic systems for general-purpose humanoid robots in the real world, and reliably benchmarking their behaviors.

Previously, I also worked on problems in natural language processing and speech and dialogue processing, with a focus on improving the algorithmic efficiency and effectiveness through mathematical principles.

Email  /  GitHub  /  Google Scholar  /  LinkedIn

Selected Research

OBCache: Optimal Brain KV Cache Pruning for Efficient Long-Context LLM Inference

We propose Optimal Brain Cache (OBCache), a principled framework that formulates KV cache eviction as a layer-wise structured pruning problem. OBCache quantifies token saliency by measuring the perturbation in attention outputs induced by pruning tokens, with closed-form scores derived for isolated keys, isolated values, and joint key-value pairs.

ESC: Efficient Speech Coding with Cross-Scale Residual Vector Quantized Transformers

We present ESC, a parameter-efficient speech foundation codec with cross-scale vector-quantized transformer architectures, which achieves coding performance comparable to state-of-the-art models while reducing decoding latency by 6.4x.

How Did We Get Here? Summarizing Conversation Dynamics

We introduce the task of summarizing the dynamics of conversations (SCD) by constructing a dataset of human-written summaries and exploring several automated baselines. We demonstrate that SCDs assist both humans and automated systems in forecasting whether an ongoing conversation will eventually derail into toxic behavior.

Design and source code from Jon Barron's website