mirror of
https://github.com/hanzoai/deepspec.git
synced 2026-09-30 21:41:22 +00:00
ci: canonical pair — hanzo.yml + forge caller
github.com has no runner for hanzo-build-linux-amd64, so a caller under .github/workflows is a gate that can never be scheduled. This is the ~7-line caller on the plane that can (git.hanzo.ai git-runner fleet), pinned @v1. Co-authored-by: Hanzo Dev <[email protected]>
This commit is contained in:
@@ -0,0 +1,15 @@
|
||||
# Canonical caller — every knob lives in /hanzo.yml, none here.
|
||||
# Runs on the git-runner fleet at git.hanzo.ai, the only pool serving these
|
||||
# labels. github.com resolves only .github/workflows and has no runner for
|
||||
# them, so a caller placed there is a gate that cannot be scheduled.
|
||||
name: CI/CD
|
||||
on:
|
||||
push:
|
||||
branches: [main]
|
||||
tags: ['v*']
|
||||
pull_request:
|
||||
workflow_dispatch:
|
||||
jobs:
|
||||
cicd:
|
||||
uses: hanzoai/ci/.hanzo/workflows/build.yml@v1
|
||||
secrets: inherit
|
||||
@@ -0,0 +1,11 @@
|
||||
# DSpark / EAGLE3 speculative-decoding training + eval. Every runnable path here
|
||||
# needs torch 2.9 + CUDA and a model checkpoint, which is a GPU node's job, not a
|
||||
# build runner's — so this gate does not pretend to train anything.
|
||||
#
|
||||
# What it does prove, over all 73 files including every config/*.py: the tree
|
||||
# still parses. That is the failure this repo actually ships — a config or a
|
||||
# trainer edited into a syntax error is invisible until a multi-hour job dies on
|
||||
# import. compileall catches it in seconds and fails the build.
|
||||
test:
|
||||
- name: compile
|
||||
run: python3 -m compileall -q -f . && echo "all python sources compile"
|
||||
Reference in New Issue
Block a user