AutoGaze automatically removes redundant patches in a video, reducing #tokens in ViT/MLLM by 4x-100x.

OpenCUA: Open Foundations for Computer-Use Agents

Official Repository of VisGym: Diverse, Customizable, Scalable Environments for Multimodal Agents

The implementation for ThreadWeaver Adaptive Threading for Efficient Parallel Reasoning in Language Models

A Large-scale Video Action Dataset

A benchmark for evaluating LLMs on open-ended CS problems. Exploring the Next Frontier of Computer Science.

Curate, Annotate, and Manage Your Data in LightlyStudio.

[NeurIPS 2025 Spotlight] Reasoning Environments for Reinforcement Learning with Verifiable Rewards

A customizable gym environment for maze/gridworld

Lifelong Learning Note

Awesome Reasoning in MLLMs: Papers and Projects about learning to reason with MLLMs, including Chain-of-Thought (CoT), OpenAl o1, and DeepSeek-R1

Our library for RL environments + evals

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

Random maze environments with different size and complexity for reinforcement learning research.

A customizable framework to create maze and gridworld environments

A framework for few-shot evaluation of language models.

A fork to add multimodal model training to open-r1

[ICCVW 25] LLaVA-MORE: A Comparative Study of LLMs and Visual Backbones for Enhanced Visual Instruction Tuning

A collection of materials for CS application

qpdf: A content-preserving PDF document transformer

A novel Multimodal Large Language Model (MLLM) architecture, designed to structurally align visual and textual embeddings.