Open Frontier Intelligence

An all-in-one, pure C++ inference engine for audio models, powered by ggml. Supports TTS, STT, VAD, voice conversion, music generation, and more, with highly optimized performance. No Python depend…

Clean, polished interface for Tencent’s SongGeneration. Create songs from text prompts or reference audio, with batch processing and smart model selection. Minimum Requirement: 10GB of VRAM

Think with AI beyond the chat box. A shared canvas for handwriting, equations, diagrams, and spatial reasoning.

Implementation of LatentMoE,Toward Optimal Accuracy per FLOP and Parameter in Mixture of Experts (Elango et al., NVIDIA 2026) — in Pytorch. A single-file, dependency-light layer you can drop in pla…

Run GLM-5.2 (744B MoE) on a 25GB-RAM consumer machine — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

Chromium fork named after radioactive element No. 90. Windows and MacOS/Raspi/Android/Special builds are in different repositories, links are towards the top of the README.md.

rvLLM — Gemma 4 inference in Rust

An optimized quantization and inference library for running LLMs locally on modern consumer-class GPUs

Data Analysis and EDA using Python

Configuration for my NixOS setup

Postgres rewritten in Rust, now passing 100% of the Postgres regression tests

Flyline: a Bash plugin to replace readline for a modern line editing experience: syntax highlighting, agent integration, rich prompts, tooltips, fuzzy history search, and more!

A tiny ~10K-parameter LLM router that learns which open-source model (deepseek-v4-pro / glm-5p2 / kimi-k2p6 via Fireworks) should answer each question and in what role, trained by evolution (sep-CM…

Build local voice agents with open-source models

common in-memory tensor structure

📚LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners🐑, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.🎉

The Web Access Gateway for AI Agents — 52 channels, 92 MCP tools, 9 SDKs, self-healing backends, async Rust

Ergonomic Framework for Humans

Learn LLM internals step by step - from tokenization to attention to inference optimization.

Fastest MoE/LLM inference runtime for consumer and edge Blackwell GPUs. SN74 on Gittensor.

slime is an LLM post-training framework for RL Scaling.

🔥 Clone and recreate any website as a modern React app in seconds

LisanBench is a lightweight benchmark for LLMs that stresses forward planning, vocabulary depth, constraint adherence, attention, and long-context "stamina" all at once.

An ahead-of-time JavaScript compiler

A PyTorch native library for training speculative decoding models

Database-Backed Durable Python Workflows