Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
Python calibrationclassificationdecision-modelhuggingfacejevmodernbertmultilingualnlppythonpytorchroutingtyped-decisionszero-shot
# DGX Spark is Linux ARM64 with NVIDIA CUDA 13.0, not Apple MPS or Jetson.
# Use instead of compose.cuda.yaml, with compose.yaml for the SDK quickstart and
# with compose.http.yaml as well for the HTTP server:
# docker compose -f compose.yaml -f compose.spark.yaml run --build --rm laya
# docker compose -f compose.yaml -f compose.http.yaml -f compose.spark.yaml up --build laya-serve
# `laya-serve` repeats the settings because it is a separate service; overrides
# for `laya` never reach it.
services:
laya:
platform: linux/arm64
build:
args:
TORCH_INDEX: "${LAYA_TORCH_INDEX:-cu130}"
environment:
LAYA_DEVICE: "${LAYA_DEVICE:-cuda}"
# Autocast dtype the runtime reads; empty keeps the checkpoint's own `amp_dtype`. Both
# services need it, because an override on `laya` never reaches `laya-serve`.
LAYA_CUDA_AMP: "${LAYA_CUDA_AMP:-}"
LAYA_CPU_AMP: "${LAYA_CPU_AMP:-}"
deploy:
resources:
reservations:
devices:
- driver: nvidia
device_ids: ["${LAYA_GPU_ID:-0}"]
capabilities: [gpu]
laya-serve:
platform: linux/arm64
build:
context: .
args:
TORCH_INDEX: "${LAYA_TORCH_INDEX:-cu130}"
environment:
LAYA_DEVICE: "${LAYA_DEVICE:-cuda}"
# Autocast dtype the runtime reads; empty keeps the checkpoint's own `amp_dtype`. Both
# services need it, because an override on `laya` never reaches `laya-serve`.
LAYA_CUDA_AMP: "${LAYA_CUDA_AMP:-}"
LAYA_CPU_AMP: "${LAYA_CPU_AMP:-}"
deploy:
resources:
reservations:
devices:
- driver: nvidia
device_ids: ["${LAYA_GPU_ID:-0}"]
capabilities: [gpu]