text-generation-inference

1
# Fork that adds only the correct stream to this kernel in order
2
# to make cuda graphs work.
3
awq_commit := bd1dc2d5254345cc76ab71894651fb821275bdd4
4

5
awq:
6
	rm -rf llm-awq
7
	git clone https://github.com/huggingface/llm-awq
8

9
build-awq: awq
10
	cd llm-awq/ && git fetch && git checkout $(awq_commit)
11
	cd llm-awq/awq/kernels && python setup.py build
12

13
install-awq: build-awq
14
	pip uninstall awq_inference_engine -y || true
15
	cd llm-awq/awq/kernels && python setup.py install
16