InternLM-XComposer2.5-OmniLive: A Comprehensive Multimodal System for Long-term Streaming Video and Audio Interactions

[ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Training

Code for the Fractured Entangled Representation Hypothesis position paper!

i2i-generator-unlearning - PyTorch - FID calculation with proper image resizing and quantization steps [CVPR 2022]

Core functionality of Kallisti Chaos Engineering Framework

The most scalable and customizable permission server on the market. Fix your slow or broken permission system with Google's proven "Zanzibar" approach. Supports ACL, RBAC, and more. Written in Go, …

LLM for Email Spam Detection

Deploy your AI/ML model to Amazon SageMaker for Real-Time Inference and Batch Transform using your own Docker container image.

CyberBench: A Multi-Task Cyber LLM Benchmark

The FL-Eval-Kit repository hosts the implementation of novel metrics introduced in our research: the Absolute Variation Distance (AVD), and the D-Rank metric.

A Multi-agent reinforcement-learning simulator framework.

GTAR l2l generator unlearning project

Chaos Engineering Framework across Private / Public / Hybrid Cloud Environments

There can be more than Notion and Miro. AFFiNE(pronounced [ə‘fain]) is a next-gen knowledge base that brings planning, sorting and creating all together. Privacy first, open-source, customizable an…

⭐️ Reason-RFT: Reinforcement Fine-Tuning for Visual Reasoning.

🤖 RoboOS: A Universal Embodied Operating System for Cross-Embodied and Multi-Robot Collaboration

[CVPR 2025] RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete. Official Repository.

[SIGGRAPH 2024] Motion I2V: Consistent and Controllable Image-to-Video Generation with Explicit Motion Modeling

The code for paper: Hierarchical Document Refinement for Long-context Retrieval-augmented Generation

New generation of CLIP with fine grained discrimination capability, ICML2025

DIAMOND (DIffusion As a Model Of eNvironment Dreams) is a reinforcement learning agent trained in a diffusion world model. NeurIPS 2024 Spotlight.

Training DIAMOND to play MarioKart64 in a Neural Network.

The Implementation of "Auto-Lambda: Disentangling Dynamic Task Relationships" [TMLR 2022].