demos: add llm #13

	name: github-actions
	on: pull_request
	jobs:
	llm_demo:
	runs-on: ubuntu-22.04
	steps:
	- uses: actions/checkout@v4
	with:
	submodules: recursive
	- name: Install OpenVINO and OpenCV
	run: \|
	mkdir ov
	curl https://storage.openvinotoolkit.org/repositories/openvino/packages/2023.1/linux/l_openvino_toolkit_ubuntu22_2023.1.0.12185.47b736f63ed_x86_64.tgz \| tar --directory ov --strip-components 1 -xz
	sudo ov/install_dependencies/install_openvino_dependencies.sh
	sudo apt install libopencv-dev
	- name: Build llm_demo
	run: \|
	mkdir build
	cd build
	cmake -DCMAKE_BUILD_TYPE=Release -DOpenVINO_DIR=../ov/runtime/cmake ..
	cmake --build . --target llm_demo --config Release -j
	- uses: actions/checkout@v4
	with:
	repository: openlm-research/open_llama_3b_v2
	ref: main
	path: open_llama_3b_v2
	lfs: true
	github-server-url: https://huggingface.co
	- uses: actions/setup-python@v4
	with:
	python-version: 3.11
	cache: 'pip'
	- name: Download and convert open_llama_3b_v2 and the vocab
	working-directory: open_llama_3b_v2
	run: \|
	git lfs checkout
	python -m pip install --extra-index-url https://download.pytorch.org/whl/cpu git+https://github.com/huggingface/optimum-intel.git
	python -c "from optimum.intel.openvino import OVModelForCausalLM; model = OVModelForCausalLM.from_pretrained('.', export=True); model.save_pretrained('.')"
	python ../demos/thirdparty/llama.cpp/convert.py . --vocab-only --outfile vocab.gguf
	- name: Run llm_demo
	run: \|
	source ov/setupvars.sh
	./build/intel64/Release/llm_demo open_llama_3b_v2/openvino_model.xml open_llama_3b_v2/vocab.gguf "return 0"

Provide feedback