Update LLM Perf Benchmarks - CUDA PyTorch #6

	name: Update LLM Perf Benchmarks - CUDA PyTorch

	on:
	workflow_dispatch:
	schedule:
	- cron: "0 0 * * *"

	concurrency:
	group: ${{ github.workflow }}-${{ github.ref }}
	cancel-in-progress: true

	jobs:
	build_image_and_run_benchmarks:
	strategy:
	fail-fast: false
	matrix:
	subset: [unquantized, bnb, awq, gptq]
	machine: [{ name: 1xA10, runs-on: [single-gpu, nvidia-gpu, a10, ci] }]

	runs-on: ${{ matrix.machine.runs-on }}

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

	- name: Build image
	run: docker build
	--build-arg USER_ID=$(id -u)
	--build-arg GROUP_ID=$(id -g)
	--build-arg TORCH_CUDA=cu121
	--build-arg CUDA_VERSION=12.1.1
	--tag llm-perf-cuda:12.1.1
	docker/cuda

	- name: Run benchmarks
	uses: addnab/docker-run-action@v3
	env:
	SUBSET: ${{ matrix.subset }}
	MACHINE: ${{ matrix.machine.name }}
	HF_TOKEN: ${{ secrets.HF_TOKEN }}
	with:
	image: llm-perf-cuda:12.1.1
	options: \|
	--rm
	--gpus all
	--shm-size 64G
	--env SUBSET
	--env MACHINE
	--env HF_TOKEN
	--env MKL_THREADING_LAYER=GNU
	--env HF_HUB_ENABLE_HF_TRANSFER=1
	--volume ${{ github.workspace }}:/workspace
	--workdir /workspace
	run: \|
	pip install packaging && pip install flash-attn einops scipy auto-gptq optimum bitsandbytes autoawq codecarbon
	pip install -U transformers huggingface_hub[hf_transfer]
	pip install -e .
	python llm_perf/update_llm_perf_cuda_pytorch.py

Provide feedback