# Laya System One

Laya System One is a self-hosted ai app. Laya is a System One decision engine. It is open source and self-hosted, installable on Docker Hub and Unraid.

Canonical page: https://ourlocalhost.com/app/2537/laya-system-one

## Description

Laya is a System One decision engine. A fast model that answers typed questions about text in one pass: pick a choice, give a score, or say yes, no or unsure, each with a calibrated probability. It serves the Jev-compatible /v1/systemone HTTP API for routing, triage, moderation and guardrails.

## Platform availability

Available on 1 of 10 platforms: Unraid.

## Health score

76/100 (Good), based on 5 of 5 factors.
- Maintained: 91 (last commit 11 days ago)
- Popular: 91 (35,507 GitHub stars)
- Easy to install: 35 (1 install platform, install notes, Docker image)
- Light to run: 30 (3 GB minimum RAM)
- Security check: 100 (no serious findings, CI, tests)

## Links

- Docker Hub: https://hub.docker.com/r/pikkonmg/laya-system-one/
- Official website: https://huggingface.co/convaiinnovations/laya
- GitHub: https://github.com/NandhaKishorM/laya
- Unraid: https://ca.unraid.net/apps/laya-system-one-1bl7j82100282e?q=laya-system-one

## Common questions

### Do I need internet access when Laya starts for the first time?

Yes. The first time the model loads, it downloads weight files from Hugging Face, so keep internet access and disk space available.

### How do I run Laya with an NVIDIA GPU?

Use a supported NVIDIA GPU and set the device to CUDA. Without it, the model may run on CPU, which is usually slower.

### How long can Laya documents be?

Laya-multilingual can read up to 8,192 tokens when max_len=8192 is passed. Accuracy beyond about 4,000 tokens of preceding text can vary, so test long documents on your own data.

### Can Laya be exposed as an HTTP service?

Yes. The laya[serve] extra provides an HTTP server, so install that extra if you want to expose Laya over HTTP.

### Can I preload Laya models instead of downloading them on first use?

The Router downloads a checkpoint on first use. If you want the models ready before requests arrive, use Router(preload=True) to load all three checkpoints up front.

---
Source: Our Local Host (https://ourlocalhost.com/app/2537/laya-system-one), the self-hosted and open-source app directory.
