8000 zigzagcai (Zheng Cai) / Starred · GitHub
[go: up one dir, main page]

Skip to content
View zigzagcai's full-sized avatar
🎯
Evolving
🎯
Evolving
  • Shanghai, China
  • 18:18 (UTC +08:00)

Block or report zigzagcai

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Showing results

FlyDSL is the Python front‑end of the project: a Flexible Layout Python DSL for expressing tiling, partitioning, data movement, and kernel structure at a high level.

Python 310 143 Updated Nov 16, 2026

Ultrafast low latency megakernel for Muse Spark Open - 1000+ tok/s

Cuda 16 1 Updated Nov 12, 2026

UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL weight transfer), and EP (e.g., GPU-driven)

C++ 1,586 198 Updated Nov 16, 2026

External kernels for miles

Cuda 2 1 Updated Nov 9, 2026

RTX 6000 Pro Wiki — Running Large LLMs (Qwen3.5-397B, Kimi-K2.5, GLM-5) on PCIe GPUs without NVLink

Python 1,206 96 Updated Nov 16, 2026

Megatron-LM patch for SpectralShift: extending Gated DeltaNet's effective context window via spectral reparameterization and learning-rate scaling.

6 Updated Sep 9, 2026
Python 326 45 Updated Nov 11, 2026

Bitwise-identical token probabilities between RL rollout (vLLM) and training

43 5 Updated Nov 8, 2026

🧱 Experimental library for building inference engines

Python 434 37 Updated Nov 16, 2026

A collection of EAT-discovered, verified GPU kernels across hardware

Cuda 5 Updated Oct 28, 2026

EAT: Evolving Agent Teams (NeurIPS'26). A supervisor agent evolves teams of coding agents for long-running optimization tasks.

Shell 130 7 Updated Nov 10, 2026

Community registry of LLM serving-path traps that produce confidently wrong measurements: templates, tool parsers, reasoning fields, quant kernel paths, CUDA toolchains, KV allocation, eval harness…

Python 178 19 Updated Nov 15, 2026

Automatic formal verification for warp-specialized GPU kernels.

C++ 112 10 Updated Nov 6, 2026

The Z3 Theorem Prover

C++ 12,943 1,732 Updated Nov 16, 2026

DeepInfer: High-performance, large-scale inference framework for DeepSeek models

Python 26,097 3,207 Updated Nov 3, 2026

Mamba SSM architecture

Python 19,007 1,843 Updated Nov 8, 2026

Making large AI models cheaper, faster and more accessible

Python 41,426 4,487 Updated Nov 16, 2026

InternEvo is an open-sourced lightweight training framework aims to support model pre-training without the need for extensive dependencies.

Python 421 67 Updated Aug 21, 2025

Public feedback discussions for: GitHub Mobile, GitHub Discussions, GitHub Codespaces, GitHub Sponsors, GitHub Issues and more!

Ruby 8,949 4,807 Updated Nov 9, 2026

Discover, compare, and contribute to cutting-edge residual connections designed for large-scale deep learning and depth scaling.

JavaScript 24 2 Updated Nov 4, 2026

State of the art inference performance with modern accelerators on Kubernetes

Python 5,280 934 Updated Nov 16, 2026

Reverse engineer anything with agents, from app behavior down to native binaries.

TypeScript 462,898 113,336 Updated Nov 16, 2026

Community maintained hardware plugin for vLLM on MetaX GPU

Python 192 85 Updated Nov 16, 2026

lithos-metal: high-performance LLM inference kernels and serving for Apple silicon.

Python 351 32 Updated Oct 21, 2026
Python 18 Updated Oct 13, 2026

🔥 The living map of LLM-driven kernel generation: 280+ agents, models, benchmarks, and industry systems across NVIDIA, AMD, Ascend, TPU, and more. Continuously updated.

Python 21 1 Updated Nov 5, 2026

Pure Rust + CUDA LLM inference engine — no PyTorch, OpenAI-compatible, serves Qwen3 to Kimi-K2

Rust 742 121 Updated Nov 14, 2026

Make videos on the computer you already own. FreeVideo runs MiniMax H3 in as little as 8 GB of VRAM and 16 GB of RAM, and adapts its acceleration path to your hardware.

Python 3,142 403 Updated Nov 14, 2026
Next
0