A python module to repair invalid JSON from LLMs

Official implementation for our paper: Repurposing AlphaFold3-like Protein Folding Models for Antibody Sequence and Structure Co-design

[NIPS 2025] Mixing Expert Knowledge: Bring Human Thoughts Back to The Game of Go. Our model is originally named InternThinker-Go, and called LoGos in our paper.

Codes for the paper "BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping" by Zhiheng Xi et al.

Official Implementation of FastMCTS: A Simple Sampling Strategy for Data Synthesis

Pre-trained, Scalable, High-performance Reward Models via Policy Discriminative Learning.

Embodied-Planner-R1: Unleashing Embodied Task Planning Ability in LLMs via Reinforcement Learning

LongLLaDA: Unlocking Long Context Capabilities in Diffusion LLMs

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

MOSS-TTSD is a spoken dialogue generation model that enables expressive dialogue speech synthesis in both Chinese and English, supporting zero-shot multi-speaker voice cloning, and long-form speech…

Scaling RL on advanced reasoning models

🚀 Efficient implementations of state-of-the-art linear attention models

An Open-source RL System from ByteDance Seed and Tsinghua AIR

Performant framework for training, analyzing and visualizing Sparse Autoencoders (SAEs) and their frontier variants.

[ICLR 2025] PEARL: Parallel Speculative Decoding with Adaptive Draft Length

DuoDecoding: Hardware-aware Heterogeneous Speculative Decoding with Dynamic Multi-Sequence Drafting

Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation

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

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

Spec-Bench: A Comprehensive Benchmark and Unified Evaluation Platform for Speculative Decoding (ACL 2024 Findings)

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

A collection of benchmarks and datasets for evaluating LLM.

[NeurIPS 2024] Can Language Models Learn to Skip Steps?

[ICLR'25] Data and code for our paper "Why Does the Effective Context Length of LLMs Fall Short?"

O1 Replication Journey

[Findings of EMNLP'2024] Unified Active Retrieval for Retrieval Augmented Generation