64 lines
2.0 KiB
YAML
64 lines
2.0 KiB
YAML
name: SSH into our runners
|
|
|
|
on:
|
|
workflow_dispatch:
|
|
inputs:
|
|
runner_type:
|
|
description: 'Type of runner to test (a10 or t4)'
|
|
required: true
|
|
docker_image:
|
|
description: 'Name of the Docker image'
|
|
required: true
|
|
num_gpus:
|
|
description: 'Type of the number of gpus to use (`single` or `multi`)'
|
|
required: true
|
|
|
|
env:
|
|
HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
|
|
HF_HOME: /mnt/cache
|
|
TRANSFORMERS_IS_CI: yes
|
|
OMP_NUM_THREADS: 8
|
|
MKL_NUM_THREADS: 8
|
|
RUN_SLOW: yes # For gated repositories, we still need to agree to share information on the Hub repo. page in order to get access. # This token is created under the bot `hf-transformers-bot`.
|
|
SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
|
|
TF_FORCE_GPU_ALLOW_GROWTH: true
|
|
CUDA_VISIBLE_DEVICES: 0,1
|
|
RUN_PT_TF_CROSS_TESTS: 1
|
|
|
|
jobs:
|
|
ssh_runner:
|
|
name: "SSH"
|
|
runs-on: ["${{ github.event.inputs.num_gpus }}-gpu", nvidia-gpu, "${{ github.event.inputs.runner_type }}", ci]
|
|
container:
|
|
image: ${{ github.event.inputs.docker_image }}
|
|
options: --gpus all --privileged --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
|
|
|
|
steps:
|
|
- name: Update clone
|
|
working-directory: /transformers
|
|
run: |
|
|
git fetch && git checkout ${{ github.sha }}
|
|
|
|
- name: Cleanup
|
|
working-directory: /transformers
|
|
run: |
|
|
rm -rf tests/__pycache__
|
|
rm -rf tests/models/__pycache__
|
|
rm -rf reports
|
|
|
|
- name: Show installed libraries and their versions
|
|
working-directory: /transformers
|
|
run: pip freeze
|
|
|
|
- name: NVIDIA-SMI
|
|
run: |
|
|
nvidia-smi
|
|
|
|
- name: Tailscale # In order to be able to SSH when a test fails
|
|
uses: huggingface/tailscale-action@main
|
|
with:
|
|
authkey: ${{ secrets.TAILSCALE_SSH_AUTHKEY }}
|
|
slackChannel: ${{ secrets.SLACK_CIFEEDBACK_CHANNEL }}
|
|
slackToken: ${{ secrets.SLACK_CIFEEDBACK_BOT_TOKEN }}
|
|
waitForSSH: true
|