Skip to content

feat: precision-driven quantization (FP16, RTN int4, static QDQ) via --precision flag #2650

feat: precision-driven quantization (FP16, RTN int4, static QDQ) via --precision flag

feat: precision-driven quantization (FP16, RTN int4, static QDQ) via --precision flag #2650

Workflow file for this run

name: Lint
on:
pull_request:
branches: [main, "release/*"]
push:
branches: [main, "release/*"]
concurrency:
group: ${{ github.workflow }}-${{ github.ref }}
cancel-in-progress: true
jobs:
lint:
runs-on: windows-latest
# Bumped from 5: combined mypy on 23 packages cold-starts at ~3-4 min on
# Windows runners; the original 5-min ceiling cancelled mid-run.
timeout-minutes: 10
steps:
- uses: actions/checkout@v4
- name: Install uv
uses: astral-sh/setup-uv@v4
- name: Set up Python
run: uv python install 3.11
- name: Install dependencies
run: uv sync --all-extras --all-groups
- name: Check license headers
run: |
uv run pre-commit run insert-license --all-files
- name: Lint
run: uv run ruff check src/ tests/
# Required type check: these packages are clean against the strict
# config in pyproject.toml. Any new mypy error here blocks the PR.
# Expand the package list as more folders are cleaned up.
#
# Single mypy invocation across all packages — a per-package loop pays
# cold typeshed/plugin startup per package and tipped the job past the
# 5-minute timeout once the list grew to 12. The combined summary still
# reports total error/file counts; error lines include file paths so
# the failing package is identifiable without per-package groups.
- name: Type check (required)
run: >-
uv run mypy
-p winml.modelkit.analyze
-p winml.modelkit.build
-p winml.modelkit.cache
-p winml.modelkit.commands
-p winml.modelkit.compiler
-p winml.modelkit.config
-p winml.modelkit.core
-p winml.modelkit.data
-p winml.modelkit.datasets
-p winml.modelkit.eval
-p winml.modelkit.export
-p winml.modelkit.inference
-p winml.modelkit.inspect
-p winml.modelkit.loader
-p winml.modelkit.onnx
-p winml.modelkit.optim
-p winml.modelkit.optracing
-p winml.modelkit.quant
-p winml.modelkit.serve
-p winml.modelkit.session
-p winml.modelkit.sysinfo
-p winml.modelkit.telemetry
-p winml.modelkit.utils