Xiaohan Yan 颜小涵
Algorithm Engineer @ AgiBot
Research interests: Multimodal Large Models, 3D Computer Vision, and Reinforcement Learning.
Feel free to drop me an email if we share any interests.
Experience
Education
Publications
* indicates equal contribution
ALOE: Action-Level Off-Policy Evaluation for Vision-Language-Action Model Post-Training
We propose ALOE, an action-level off-policy evaluation framework for VLA post-training that enables fine-grained credit assignment and stable policy improvement in real-world robotic manipulation.
HOLO: Holistic Lightweight Optimization for Scene Understanding with Auto-Annotation and Multimodal Learning
We propose HOLO, which includes a large-scale scene description dataset and a lightweight 3D-LLM.
RE0: Recognize Everything with 3D Zero-shot Instance Segmentation
Given 3D point clouds and multi-view RGB-D images with poses, RE0 leverages the 3D geometric information, projection relationships and CLIP semantic features for 3D zero-shot instance segmentation.
Semantic-Guided Gaussian Splatting with Deferred Rendering
We use semantic features derived from 2D foundation model to revolutionize the material property optimization for 3DGS.
AttenPoint: Exploring Point Cloud Segmentation through Attention-Based Modules
We combine local and global features of the structures and performance to perform few-shot point cloud semantic segmentation.
GreedyAgent: A Simple yet Efficient Approach for Meta Learning from Learning Curves
A key sub-problem: meta-learning from learning curves is a mature but gradually gaining attention area within the field of meta-learning.
Anatomical Structure-Guided Medical Vision-Language Pre-training
We propose an anatomical structure-guided framework for medical vision-language pre-training that improves cross-modal alignment.
Projects
LLM Science Exam — Use LLMs to Answer Difficult Science Questions
Fine-tuning large language models on private datasets. Score 0.905, ranking top 3% worldwide, Silver Medal.
eScape — A Geometry Storm Game
Scale your device and escape from this geometry storm. Innovation RK1 and Theme interpretation RK2.
Awards
Most of the awards I won during my student years.
2018 – 2024
The 2019 ICPC Asia-East Continent Final — Bronze Medal
2019 – 2020
Jiangsu Collegiate Programming Contest — Silver Medal 2nd place
2019 – 2020
News
Misc
3D Modeling
- Some stuff I have modeled and rendered.
Sports
- Swimming
- Flying Disc
Games
- Pokemon
- Digimon
Language
- Japanese