Ge Zhang

A new system that connects music and language for better understanding.

AnyGPT processes text, images, music, and speech for diverse applications.

The Yi model family showcases strong language and multimodal processing capabilities.

Discover the processes behind training advanced AI language models.

MuPT utilizes ABC notation for effective music generation with AI.

MAP-Neo aims for transparency and performance in AI language modeling.

MMLU-Pro challenges language models with harder questions and more answer options.

A new method helps balance training data for better AI performance.

Examining the difficulties of translating metaphorical language in machine translation.

GIEBench assesses LLMs' empathetic responses based on diverse group identities.

DDK enhances knowledge distillation, making smaller language models more efficient.

I-SHEEP enables large language models to learn continuously from generated data.

Exploring how preference learning improves language model alignment with human expectations.

A new tool evaluates large language models' performance across multiple data types.

Researchers use gaming glitches to teach AI about physical commonsense.