Sun Haoyuan_Thu

MIO. Python

Mask-GRPO. The code for NeurIPS2025 paper "Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image Generation"

Awesome-RL-based-Reasoning-MLLMs. This repository provides valuable reference for researchers in the field of multimodality, please start your exploratory travel in RL-based Reasoning MLLMs!

AIGC-Interview-Book. 【三年面试五年模拟】AIGC/LLM/AI Agent算法工程师面试秘籍。涵盖AIGC、LLM大模型、AI Agent、传统深度学习、自动驾驶、机器学习、计算机视觉、自然语言处理、强化学习、大数据挖掘、具身智能、元宇宙、AGI等AI行业面试笔试干货经验与核心知识。

R1-Omni. Python

UAV-DETR. Python

Skywork-R1V. Skywork-R1V is an advanced multimodal AI model series developed by Skywork AI, specializing in vision-language reasoning.

Awesome_MLLMs_Reasoning.

MM-EUREKA. MM-EUREKA: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning

VisualThinker-R1-Zero. Explore the Multimodal “Aha Moment” on 2B Model

UAVs_Meet_LLMs.

EasyR1. EasyR1: An Efficient, Scalable, Multi-Modality RL Training Framework based on veRL

Open-R1-Video. ✨First Open-Source R1-like Video-LLM [2025/02/18]

R1-V. Witness the aha moment of VLM with less than $3.

TinyZero. Minimal reproduction of DeepSeek R1-Zero

VLM-R1. Solve Visual Understanding with Reinforced VLMs

LlamaV-o1. [ACL 2025 🔥] Rethinking Step-by-step Visual Reasoning in LLMs

class-balanced-loss. Class-Balanced Loss Based on Effective Number of Samples. CVPR 2019

eurosat_classification. Land Cover Classificatio using the EuroSAT dataset

Multimodal-AND-Large-Language-Models. Paper list about multimodal and large language models, only used to record papers I read in the daily arxiv for personal needs.

Vision-LLM-Alignment. This repository contains the code for SFT, RLHF, and DPO, designed for vision-based LLMs, including the LLaVA models and the LLaMA-3.2-vision models.

safe-sora. SafeSora is a human preference dataset designed to support safety alignment research in the text-to-video generation field, aiming to enhance the helpfulness and harmlessness of Large Vision Models (LVMs).

VADER. Video Diffusion Alignment via Reward Gradients. We improve a variety of video diffusion models such as VideoCrafter, OpenSora, ModelScope and StableVideoDiffusion by finetuning them using various reward models such as HPS, PickScore, VideoMAE, VJEPA, YOLO, Aesthetics etc.

Baton. Official Repository of IJCAI 2024 Paper: "BATON: Aligning Text-to-Audio Model with Human Preference Feedback"

EAFO-code. EAFO-code for ICLR2025

DNRT. [ICLR 2024] Dynamic Neural Response Tuning

Decision-ConvFormer. [ICLR 2024 Spotlight] Code for the paper "Decision ConvFormer: Local Filtering in MetaFormer is Sufficient for Decision Making"

Expected-Calibration-Error. Python

direct-preference-optimization. Reference implementation for DPO (Direct Preference Optimization)

MBPO-pytorch. Model-based Policy Optimization re-implement by pytorch

ms-agent. MS-Agent: a lightweight framework to empower agentic execution of complex tasks

d2l-zh. 《动手学深度学习》：面向中文读者、能运行、可讨论。中英文版被70多个国家的500多所大学用于教学。