An agentic skills framework & software development methodology that works.

Convert arxiv papers to markdown

Agent skills for Obsidian. Teach your agent to use Markdown, Bases, JSON Canvas, and use the CLI.

Multi-agent reconnaissance skill for Claude Code + Obsidian

微信公众号转RSS

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models

FlashMLA: Efficient Multi-head Latent Attention Kernels

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

DeepEP: an efficient expert-parallel communication library

Magnum IO community repo

Train transformer language models with reinforcement learning.

ByteCheckpoint: An Unified Checkpointing Library for LFMs

An Easy-to-use, Scalable and High-performance Agentic RL Framework based on Ray (PPO & DAPO & REINFORCE++ & VLM & TIS & vLLM & Ray & Async RL)

verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework

Infiniband Verbs Performance Tests

RDMA core userspace libraries and daemons

Large Context Attention

🚀 A simple way to launch, train, and use PyTorch models on almost any device and distributed configuration, automatic mixed precision (including fp8), and easy-to-configure FSDP and DeepSpeed support

🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch.

📚A curated list of Awesome LLM/VLM Inference Papers with Codes: Flash-Attention, Paged-Attention, WINT8/4, Parallelism, etc.🎉

A PyTorch native platform for training generative AI models

DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model

Rotary Transformer

Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.

Byted PyTorch Distributed for Hyperscale Training of LLMs and RLs

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

深度学习经典、新论文逐段精读

Ongoing research training transformer language models at scale, including: BERT & GPT-2

Emote Portrait Alive: Generating Expressive Portrait Videos with Audio2Video Diffusion Model under Weak Conditions

A QoS-based scheduling system brings optimal layout and status to workloads such as microservices, web services, big data jobs, AI jobs, etc.