AI Infra2025
Spins up sandboxed VMs and containers (macOS, Linux, Windows, Android) that an AI agent can fully control through one unified SDK, cloud or local, plus a benchmark suite and background drivers that automate native apps without grabbing the cursor.
AI Infra2025
High-performance CUDA tensor-core GEMM kernel library for LLM workloads: supports FP8/FP4/BF16, fused Mega MoE and MQA scoring, and runtime JIT-compiled kernels. Targets NVIDIA SM90/SM100 and PyTorch—for teams working on low-level GPU kernel optimization.
AI Infra2025
Provides high-throughput, low-latency GPU communication kernels for Mixture-of-Experts (MoE) and expert-parallel workloads, with NVLink↔RDMA-aware forwarding, FP8/BF16 support, and low-latency RDMA hooks for inference decoding.
AI Train2025
An asynchronous, high-throughput framework for large-scale reinforcement learning and agentic training that scales to 1T+ MoE models and 1000+ GPUs, with native verifiers integration, end-to-end SFT/RL/evals, and Slurm/Kubernetes deployment; requires NVIDIA GPUs.
AI Train2025
inclusionAI (AReaL Team), Institute for Interdisciplinary Information Sciences, Tsinghua University +1Ant Group, Institute for Interdisciplinary Information Sciences, Tsinghua University
Trains LLM reasoning and agentic models with fully asynchronous reinforcement learning, decoupling rollout generation from policy updates for a 2.77x speedup over synchronous RL. Covers GRPO, PPO and DAPO across Megatron, FSDP, vLLM and SGLang backends.
AI Dataset2025
High-quality, efficiently verified and filtered web corpus for LLM pretraining — supplies ~1 trillion English tokens and ~120 billion Chinese tokens with English/Chinese Parquet splits. Designed for large-scale pretraining experiments and data-filtering research.
Embodied AI2025
A vision-language-action foundation model and reference stack for generalized humanoid and cross-embodiment robot manipulation. Provides pretrained checkpoints, demo datasets, and tooling for fine-tuning, evaluation, and deployment (ONNX/TensorRT); released as Early Access.
AI Infra2025
Runs and fine-tunes LLMs locally on Apple silicon via the MLX framework, pulling thousands of Hugging Face models with one command. Adds 4- and 8-bit quantization, LoRA and full fine-tuning, prompt caching, and distributed inference across Macs.
AI Train2025
Trains multi-step LLM agents with reinforcement learning (GRPO) on your own tasks, wrapping existing agent code behind an OpenAI-compatible client. Its RULER mode scores trajectories with an LLM judge, so there's no reward function to hand-write.
AI Image2025
Real-time DETR detector on a DINOv2 backbone, covering detection, segmentation, and keypoints. Ships in six sizes (Nano to 2XL), beats YOLO on the COCO speed-accuracy curve, and transfers better to non-COCO real-world domains.
AI Dataset2025
Provides 5 million instruction–response pairs for supervised fine-tuning of code LLMs, with inputs, outputs, unit tests, and automated LLM judgments. Uses hybrid automated/synthetic generation and is released under CC BY 4.0 for large-scale SFT workflows.
AI Train2025
Provides a modular full-stack reinforcement learning stack to train and evaluate long-horizon, multi-turn tool-use LLM agents, including a performant trainer, a Tinker-compatible backend, agent orchestration, and Gymnasium-style environments for task design.