arxiv:2605.14040
Shan Yang
shanyangmie
AI & ML interests
Multimodal reasoning · post-training (SFT/DPO/RL) · physics-grounded VLMs.
Staff Applied Scientist @ Adobe Foundry. PhD UNC (advisor: Ming C. Lin). Prev: Google Research, Amazon.
Recent: [Physics-R1](https://arxiv.org/abs/2605.14040) — +18.9 pp on PhysOlym-A via RL post-training.
Co-author on Attention Bottlenecks for Multimodal Fusion (NeurIPS 2021), AI Choreographer / AIST++ (ICCV 2021), ICAR (AAAI 2024), VLAP (ECCV 2024).
Recent Activity
new activity about 14 hours ago
shanyangmie/physr1corp:missing image files updated a dataset about 14 hours ago
shanyangmie/physr1corp updated a model 4 days ago
shanyangmie/physics-r1-seed17Organizations
None yet
models 21
shanyangmie/physics-r1-seed17
Image-Text-to-Text • 9B • Updated • 28
shanyangmie/physics-r1-seed23
Image-Text-to-Text • 9B • Updated • 30
shanyangmie/physics-r1-seed42-v4-step60
Image-Text-to-Text • 9B • Updated • 32
shanyangmie/physics-r1-seed17-v4-step40-fsdp
Image-Text-to-Text • Updated
shanyangmie/physics-r1-seed17-v4-step50-fsdp
Image-Text-to-Text • Updated
shanyangmie/physics-r1-seed42-v4-step40-fsdp
Image-Text-to-Text • Updated
shanyangmie/physics-r1-seed42-v4-step50-fsdp
Image-Text-to-Text • Updated
shanyangmie/physics-r1-seed42-v4-step60-fsdp
Image-Text-to-Text • Updated
shanyangmie/physics-r1-seed23-canonical-step60-fsdp
Image-Text-to-Text • Updated
shanyangmie/physics-r1-seed17-canonical-step63-fsdp
Image-Text-to-Text • Updated
datasets 6
shanyangmie/physr1corp
Viewer • Updated • 2.27k • 82
shanyangmie/physcorp-pre-audit
Viewer • Updated • 14.3k • 79
shanyangmie/physolym-a
Updated • 78
shanyangmie/physcorp-a
Viewer • Updated • 6.43k • 113 • 1
shanyangmie/physics-r1-eval-outputs
Preview • Updated • 191
shanyangmie/physdojo-annotations
Viewer • Updated • 11.6k • 90