mizorewww/laya-coreml
Local Laya typed decisions on Apple Core ML and Neural Engine. Validated ports, ~5 ms short decisions on M3 Max, reproducible speed and energy benchmarks.
From the publisher
README & documentation
Source previewRead the project’s overview, installation instructions and usage examples. The original README is the source of truth.
Open-weight typed decisions on Apple Silicon. Core ML, Neural Engine, zero generated tokens. PyPI · Hugging Face weights · 中文 A real Laya model plays Snake locally, with visible probabilities, score, length, latency and safety interventions. The GIF replays a recorded Core ML run at 1× speed. The game uses explicit planner features and a visible cycle safety layer. The complete active Snake loop sustained 49.1–50.0 decisions/s across three uncapped 600-step episodes, with zero deaths and two safety interventions. Game-loop timings and paced-rate limits include rendering serialization; terminal painting is excluded. One short multilingual decision: 4.98 ms P50 / 5.31 ms P95 on M3 Max with ANE FP16. The same experiment measured 2.78× better whole-system energy per decision than compiled MLX FP16. A separately validated W8 palette variant reached 4.88 ms and 3.19× energy improvement. These are single-question results, not full Snake frame times; the requested 10× improvement was not…
Read the full README ↗ · Preview checked 2026-09-21T07:41:40.320Z
Inside the original README — Document outline
- Laya-CoreML
- Run the demo
- Ask for a decision
- Measured on M3 Max
- Available checkpoints
- Port fidelity and limits
- Documentation and reproducibility
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Links from the README
References supplied by the publisher, not independently verified endorsements. Check the destination before downloading files or entering credentials.
- 中文
- Game-loop timings and paced-rate limits
- Controls, recording and video export
- Shareable video and recording provenance
- Full API, model selection and offline usage
- Speed, energy and hardware evidence
- Release artifacts and pinned Hub revisions
- ANE engineering experiments
- Mathematical investigation of the 10× target
- Conversion issues and supported shapes
Documentation belongs to its respective authors. Reported project/model license: Apache-2.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
What this repository does
Local Laya typed decisions on Apple Core ML and Neural Engine. Validated ports, ~5 ms short decisions on M3 Max, reproducible speed and energy benchmarks.
Repository facts
- Owner
- mizorewww
- Primary language
- Python
- Stars
- 486
- Forks
- 28
- Open issues + pull requests
- 0
- License
- Apache-2.0
- Archived
- No
- Default branch
- main
- Created
- 2026-09-19T23:14:13.000Z
- Last push
- 2026-09-20T03:52:14.000Z
Topics and intended use
Owner-supplied topics: apple-neural-engine, apple-silicon, coreml, decision-model, laya, local-ai, modernbert, on-device-ai, typed-decisions
Review the README for scope, installation, examples and limitations. We do not run repository code or certify it.
Evaluate before installing
Review licensing and dependencies, inspect recent commits and unresolved issues, and test in an isolated environment before production use. Stars and forks alone cannot answer these questions.
Source and freshness
Source: GitHub. Metadata observed 2026-09-21T06:41:42.808Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.
Related repositories
EbookFoundation/free-programming-books
donnemartin/system-design-primer