From 921b92a69f36502dc5eefd14ba3487c1bb49bb9d Mon Sep 17 00:00:00 2001 From: yum Date: Wed, 13 Dec 2023 13:54:55 -0800 Subject: Begin experimenting with flash-attention Seems much faster than faster-whisper. There are two issues: * Requires NVIDIA 3000 series or higher. * Incompatible with faster-whisper dependencies. So it seems like we'll either need to toggle between two sets of dependencies at runtime or have two environments. --- Scripts/requirements.txt | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'Scripts/requirements.txt') diff --git a/Scripts/requirements.txt b/Scripts/requirements.txt index 9224ba8..58db539 100644 --- a/Scripts/requirements.txt +++ b/Scripts/requirements.txt @@ -1,17 +1,20 @@ -ctranslate2 editdistance -faster-whisper@https://github.com/guillaumekln/faster-whisper/archive/78d57d73c5b4a76b32d1d5a415e4e7aea760295c.tar.gz +flash-attn==2.3.6 future==0.18.2 huggingface_hub==0.16.4 keyboard langcodes language-data openvr +onnxruntime +packaging pillow pyaudio pydub python-osc pyyaml sentence_splitter -transformers>=4.21.0 +transformers==4.35.2 +--extra-index-url https://download.pytorch.org/whl/cu121 +torch -- cgit v1.2.3