Yang Chen
Research Scientist at NVIDIA (ADLR).
Currently
I work on LLM post-training building towards self-improving AI.
Previously Ph.D. at Georgia Tech (2024), with research on multimodal and
multilingual language models.
§ 01 — Selected Projects
Recent research.
A few projects I've led or contributed to at NVIDIA ADLR. Most ship as open models
with training data released on Hugging Face.
§ 02 — Past Work
Earlier threads
from 2019–2024.
Research directions I pursued during my Ph.D. at Georgia Tech — multimodal understanding
and cross-lingual representation for language models.
2022 — 202404 papers
Multimodal LLM
Building AI that understands the visual world — world knowledge, privacy, and universal multimodal retrieval.
- Visual World Knowledge EMNLP'23 ICCV'23 Oral
- Emergent Visual Privacy EMNLP'24 Oral
- Universal Multimodal Retriever ECCV'24 Oral