Back to directory

Laya-CoreML

A Core ML and Apple Neural Engine port of the Laya typed-decision model, which answers structured choice, ordinal-score and yes/no questions in a few milliseconds without generating text tokens.

Laya-CoreML

Laya-CoreML runs the open-weight Laya typed-decision model locally on Apple Silicon through Core ML and the Neural Engine. Instead of generating text tokens, it returns probabilities for a defined choice, an ordinal score, or a boolean answer to a structured question, such as whether a customer is requesting a refund, which the project reports at close to 5 milliseconds per short decision on an M3 Max.

A bundled demo plays Snake locally using the model's decisions, with visible probabilities, score and safety interventions, to show the loop running live. Models download once from Hugging Face and then run fully offline; it installs with pip install laya-coreml and requires macOS 15 or later with Python 3.11 to 3.13.

More in Models

See all 405
  • mini-AGI

    github.com

    A continual learning language model that trains from scratch on a single 8 GB VRAM GPU, assembling its own architecture and adapting its size to the hardware and data available rather than shipping as a fixed pretrained model.

  • An open-source contrastive-embedding pipeline that maps technical CVE vulnerability descriptions to formal cybersecurity compliance controls in NIST SP 800-53 and CMMC, published with its model generation, fine-tuning and evaluation code.

  • peekaboolean

    github.com

    A small vision-language model, fine-tuned with LoRA on a 500-million-parameter backbone, that answers typed multiple-choice, ordinal-score and yes/no questions about an image with calibrated probabilities, in about 400 milliseconds on a laptop.