Fast and memory-efficient exact attention
Repository Index
132,241 visible repositories in the index
1–10 of 10 · Most stars
The fastest way to run Qwen 3.8 Flash Next, Qwen 3.8 27B and Ternary Bonsai 2 27B on a Mac: 125 tok/s in OpenCode on an M5 Max, and a 27B model on 16 GB Macs. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anthropic compatible local server.
- apple-silicon
- local-ai
- mlx
- mtp
Verify that the PowerPoint you are about to deliver is exactly the PPTX that passed QA—locally, offline, and with tamper-evident evidence.
- artifact-integrity
- pptx
- python
- supply-chain-security
Fast and memory-efficient exact attention
A cyber brain for your AI. It never forgets a detail, remembers exactly what you said, and learns how you work over time. Free, local, works with Cursor, Claude Code, Codex, OpenClaw, Hermes and more. MIT.
A computer, compiled into the weights of a language model. An exact CPU inside Qwen3.5-4B, with Gated DeltaNet state as RAM.
Fast, exact inference for StartLux-Decision models on NVIDIA GPUs: the released GGUF files on int8 and 4-bit tensor cores, behind the /v1/systemone API, with up to 5.3× lower latency than vLLM in FP8 and bit-identical answers.
- calibration
- cuda
- cuda-graphs
- decision-making
Cut AI context cost without trusting the compressor. Every reduction is reversible, byte-exact recoverable, and carries an auditable receipt. Local-first, works through proxy, MCP, SDK, or agent wrapper.
- ai-agents
- claude-code
- mcp
- token-optimization
Claude Code plugin: game sprites at your game's exact camera angle, drawn in your art style - props, 8-direction walking characters, characters seated on your own chairs (3D-guided)
Security linter for AI-generated code: catches the exact vulnerabilities Copilot, Cursor and ChatGPT repeatedly produce. SQL injection, hardcoded secrets, JWT bypass, command injection and 15+ more rules. Grade A–F. Zero config. CI/CD ready.