Official repo of VLABench, a large scale benchmark designed for fairly evaluating VLA, Embodied Agent, and VLMs.

[ICLR2025] ReAttention, a training-free approach to break the maximum context length in length extrapolation

a survey of long-context LLMs from four perspectives, architecture, infrastructure, training, and evaluation

For OpenMOSS Mechanistic Interpretability Team's Sparse Autoencoder (SAE) research.

Ultra-Innerthought is a bilingual (Chinese and English) open-domain R1/o1 style SFT dataset.

GPT-4o-level, real-time spoken dialogue system.

Collaborative Training of Large Language Models in an Efficient Way

Code for "AnyGPT: Unified Multimodal LLM with Discrete Sequence Modeling"