text-generation-inference

1
flash_att_commit := 3a9bfd076f98746c73362328958dbc68d145fbec
2

3
flash-attention:
4
    # Clone flash attention
5
	pip install -U packaging ninja  --no-cache-dir
6
	git clone https://github.com/HazyResearch/flash-attention.git
7

8
build-flash-attention: flash-attention
9
	cd flash-attention && git fetch && git checkout $(flash_att_commit)
10
	cd flash-attention && python setup.py build
11
	cd flash-attention/csrc/rotary && python setup.py build
12
	cd flash-attention/csrc/layer_norm && python setup.py build
13

14
install-flash-attention: build-flash-attention
15
	pip uninstall flash_attn rotary_emb dropout_layer_norm -y || true
16
	cd flash-attention && python setup.py install && cd csrc/layer_norm && python setup.py install && cd ../rotary && python setup.py install
17