Add `vllm` decoder for model inference (#124) #29

	name: Format leaderboard

	on:
	push:
	branches:
	- main
	paths:
	- 'src/alpaca_eval/leaderboards/data_AlpacaEval/**'
	- 'results/**'
	workflow_dispatch:

	jobs:
	format_leaderboard:
	runs-on: "ubuntu-20.04"

	steps:
	- name: Checkout code
	uses: actions/checkout@v3

	- name: Set up Python
	uses: actions/setup-python@v4
	with:
	python-version: '3.10'
	cache: "pip"

	- name: Install dependencies
	run: pip install .

	- name: Format sample sheets
	run: python docs/format_sample_sheets.py

	- name: Update leaderboard
	run: python docs/format_export_leaderboards.py

	- name: Configure Git
	run: \|
	git config user.name "GitHub Actions"
	git config user.email "[email protected]"

	- name: Commit and push if changed
	run: \|
	git diff
	git diff --quiet \|\| (git add . && git commit -m 'Automated leaderboard update')
	git push https://${{secrets.GITHUB_TOKEN}}@github.com/tatsu-lab/alpaca_eval.git HEAD:main
	env:
	GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Provide feedback