Papers


Selected Works

TALK SLIDES

Towards Better Global landscape of GANs: How 2 Lines of Code Change Makes Difference [slides]

Over-parameterized networks have no bad basins [slides]

When Do Neural Networks Have No Bad Local Minima? (for ICML and NeurIPS’19 papers) [slides]

PREPRINTS

Recent works on large language models:

Some earlier works:

PUBLICATIONS (by Time)

$^+$: co-first author; $^{*}$: corresponding author

PATENTS

Ph.D. Dissertation

Advised master projects (selected)