Skip to content

get_max_sequence_length() warning if user MAX_SEQUENCE_LENGTH > model MAX_SEQUENCE_LENGTH #386

get_max_sequence_length() warning if user MAX_SEQUENCE_LENGTH > model MAX_SEQUENCE_LENGTH

get_max_sequence_length() warning if user MAX_SEQUENCE_LENGTH > model MAX_SEQUENCE_LENGTH #386

Workflow file for this run

name: "Test"
on:
workflow_dispatch:
push:
branches:
- main
paths-ignore:
- "**.md"
- "proto/**"
pull_request:
branches:
- main
paths-ignore:
- "**.md"
- "proto/**"
defaults:
run:
shell: bash
env:
CI: true
TEST_IMAGE_NAME: "cpu-tests:0"
jobs:
build:
runs-on: ubuntu-latest
permissions:
packages: write
contents: read
env:
CACHE_IMAGE: "ghcr.io/ibm/text-gen-server:test-cache"
CACHE_REGISTRY: "ghcr.io"
steps:
- name: "Checkout"
uses: actions/checkout@v4
- name: "Set up QEMU"
uses: docker/setup-qemu-action@v3
- name: "Setup Docker Buildx"
uses: docker/setup-buildx-action@v3
- name: "Log in to cache image container registry"
uses: docker/login-action@v3
if: github.event_name != 'pull_request'
with:
registry: ${{ env.CACHE_REGISTRY }}
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: "Set build cache target"
run: |
# For push to `main` (PR merged), push a new cache image with all layers (cache-mode=max).
# For PR builds, use GitHub action cache which isolates cached layers by PR/branch.
# to optimize builds for subsequent pushes to the same PR/branch.
# Do not set a cache-to image for PR builds to not overwrite the `main` cache image and
# to not ping-pong cache images for two or more different PRs.
# Do not push cache images for each PR or multiple branches to not exceed GitHub package
# usage and traffic limitations.
# UPDATE 2024/02/26: GHA cache appears to have issues, cannot use `cache-to: gha,mode=min`
# if `cache-from: reg...,mode=max` but `cache-to: gha,mode=max` takes longer than uncached
# build and exhausts GHA cache size limits, so use cache `type=inline` (no external cache).
if [ "${{ github.event_name }}" == "pull_request" ]
then
#CACHE_TO="type=gha,mode=min"
CACHE_TO="type=inline"
else
CACHE_TO="type=registry,ref=${{ env.CACHE_IMAGE }},mode=max"
fi
echo "CACHE_TO=$CACHE_TO" >> $GITHUB_ENV
- name: "Build test image"
uses: docker/build-push-action@v5
with:
context: .
target: "cpu-tests"
tags: ${{ env.TEST_IMAGE_NAME }}
cache-from: |
type=gha
type=registry,ref=${{ env.CACHE_IMAGE }}
cache-to: ${{ env.CACHE_TO }}
outputs: type=docker,dest=/tmp/test_image.tar
- name: "Upload test image"
uses: actions/upload-artifact@v4
with:
name: "test-image"
path: /tmp/test_image.tar
retention-days: 1
test-python:
runs-on: ubuntu-latest
needs: build
steps:
- name: "Checkout"
uses: actions/checkout@v4
- name: "Download test image"
uses: actions/download-artifact@v4
with:
name: "test-image"
path: /tmp
- name: "Load Docker image"
run: |
docker load --input /tmp/test_image.tar
docker image ls -a
- name: "Run Python tests"
run: |
make python-tests
integration-tests:
runs-on: ubuntu-latest
needs: build
steps:
- name: "Checkout"
uses: actions/checkout@v4
- name: "Download test image"
uses: actions/download-artifact@v4
with:
name: "test-image"
path: /tmp
- name: "Load Docker image"
run: |
docker load --input /tmp/test_image.tar
docker image ls -a
- name: "Run integration tests"
run: make integration-tests