neurips-2025 开源项目推荐（79 个）

taco-group /4KAgent

[NeurIPS 2025] 4KAgent：将任何图像扩展到4K超分辨率。一款智能计算机视觉代理，可以神奇地将任何图像恢复到完美-4K！

mll-lab-nu /VAGEN

World model reinforcement learning for multi-turn VLM agents. RL for vision framework (NeurIPS 2025).

VCIP-RGBD /DFormer

[CVPR 2025]DFormerv2：RGBD语义分割的几何自关注 & [ICLR 2024]DFormer 与 [NeuriPS 2025] OmniSegmentor

ziangcao0312 /PhysX-3D

PhysX: Physical-Grounded 3D Asset Generation (NeurIPS 2025, Spotlight)

MadeAgents /mobile-use

MobileUse：一款开源的移动图形界面代理，用于Android手机自动化、AndroidWorld/AndroidLab评估、层级反思和主动探索。

doujiang-zheng /Awesome-Graph-Learning-Papers-List

Advances on machine learning of graphs, covering the reading list of recent top academic conferences.

FastMAS /KVCOMM

[NeurIPS'25] KVCOMM: Online Cross-context KV-cache Communication for Efficient LLM-based Multi-agent Systems

ForeverBlue816 /PhysioWave

[NeurIPS 2025] PhysioWave 与 PhysioWave-T：面向 EEG、ECG 和 sEMG 的令牌高效小波基础模型，具备可靠性感知的多模态融合。

ziplab /ZPressor

[NeurIPS 2025] ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGS

mims-harvard /CUREBench

CUREBench @ NeurIPS 2025: Benchmarking AI reasoning for therapeutic decision-making at scale

aim-uofa /Omni-R1

[NeurIPS 2025] Official Repo of Omni-R1: Reinforcement Learning for Omnimodal Reasoning via Two-System Collaboration

Linketic /TC-Light

[NeurIPS`25] TC-Light: Temporally Coherent Generative Rendering for Realistic World Transfer

lck666666 /plana3r

[NeurIPS 2025] the official project page of a paper, "PLANA3R: Zero-shot Metric Planar 3D Reconstruction via Feed-Forward Planar Splatting"

doujiang-zheng /Dynamic-Temporal-Graph-Learning-Reading-List

Advances on machine learning of dynamic (temporal) graphs, covering the reading list of recent top academic conferences.

dqj5182 /HACO_RELEASE

[NeurIPS 2025] This repo is official PyTorch implementation of the paper "Learning Dense Hand Contact Estimation from Imbalanced Data".

IRVLUTD /HO-Cap

[NeurIPS 25'] A Python package that provides evaluation and visualization tools for the HO-Cap dataset

Mobiuslqm /Nabla-R2D3

[NeurIPS 2025 Official Codes] Nabla-R2D3: Effective and Efficient 3D Diffusion Alignment with 2D Rewards

HewlettPackard /sustain-lc

Sustain-LC is a benchmarking environment for traditional and reinforcement learning based controls as well as LLM based control

apanariello4 /core-space-merging

Pytorch code for NeurIPS 2025 paper "Accurate and Efficient Low-Rank Model Merging in Core Space"

mlvlab /DeepVideoR1

[NeurIPS25] Official Implementation (Pytorch) of "DeepVideo-R1"