Gaowen Liu

Critique-aware training can boost LLM agent reliability by over 10% in complex, stateful environments, transforming how we approach tool-calling tasks.

VLM-level feedback can transform language backbone unlearning from unreliable to robust and transferable, achieving unprecedented performance gains.

CLIP's latent space is better understood as a hyperspherical semantic mixture, revealing richer and more interpretable representations than traditional Gaussian models.