Tag
Explore by tags
Self-hosted coding assistant that runs frozen local LLMs with constraint-driven planning, energy-based verification, and self-verified repair to produce verified code. Emphasizes offline inference (no cloud), Docker/bare-metal deployment, and requires a 16GB+ GPU.
Multimodal OCR and document-understanding toolkit for recognizing complex layouts, tables, formulas and code. Uses Multi-Token Prediction and stable RL for better training; ships as a 0.9B-parameter model with a Python SDK and deployment guides for vLLM, SGLang and Ollama.
Enables research-grade character animation with neural networks in a single NumPy/PyTorch environment — train models, run inference, and visualize results without leaving Python. Includes ECS-style architecture, mocap import (GLB/FBX/BVH), built-in renderer, and headless/standalone modes for rapid prototyping.
Runs untrusted code (LLM outputs, plugins, and third‑party tools) inside cross‑platform, policy‑driven sandboxes. Provides a unified JSON schema and a TypeScript SDK that sit on multiple containment backends (process sandboxes, LXC/Bubblewrap, microVMs). Early preview with known permissive profiles — not yet a security boundary.
Runs a local-first, full AI stack—LLM inference, chat UI, voice, agents, workflows, RAG, and image generation—deployable with one command. Auto-detects hardware and bootstraps a small model for instant chat while larger models download; supports Linux, Windows, macOS and optional cloud/hybrid modes.
Turns a PC, Mac, or Linux machine into a private AI server with one-command installers: local LLM inference, a ChatGPT-style web UI, voice, agents, RAG, workflows, image generation, hardware-aware model selection, and optional cloud/hybrid modes.
Local LLM inference server for Apple Silicon that exposes an OpenAI-compatible API and a macOS menubar app. Uses continuous batching and a two-tier KV cache (RAM + SSD in safetensors) to persist context across restarts, enabling practical multi-model serving and fast local coding workflows.
Acts as an OpenAI‑compatible local and cloud gateway that routes requests across 100+ LLM providers with smart routing, load balancing, retries and fallbacks. Adds policies, rate limits, semantic caching and observability for reliable, cost‑aware inference in Docker, Electron or npm installs.
Provides a reliability layer for self-hosted LLM tool-calling and multi-step agent workflows. Adds guardrails — rescue parsing, response validation, retry nudges, and a synthetic respond tool — and ships a Drop-in OpenAI-compatible proxy plus a WorkflowRunner for structured loops.
Runs local AI models on Apple Silicon as an OpenAI‑compatible server, emphasizing low latency, prompt caching, and reliable tool-calling. Optimized for M1–M4 Macs with multimodal support and drop‑in compatibility for IDEs and agent frameworks.
Runs transformer forward+backward training directly on Apple's Neural Engine by reverse-engineering private ANE APIs; includes per-layer ANE kernels, INT8 optimizations and benchmarks. Proof-of-concept only—relies on undocumented APIs, has low utilization and CPU fallbacks.