Bump to flash-attention 2.3.4 + switch to Github Actions wheels on Windows (#4700)

This commit is contained in:
oobabooga 2023-11-21 20:06:56 -03:00
parent b81d6ad8a4
commit fb124ab6e2
3 changed files with 17 additions and 13 deletions

View file

@ -303,7 +303,7 @@ def update_requirements(initial_installation=False):
elif is_cuda118:
textgen_requirements = [req.replace('+cu121', '+cu118').replace('+cu122', '+cu118') for req in textgen_requirements]
if is_windows() and (is_cuda117 or is_cuda118): # No flash-attention on Windows for CUDA 11
textgen_requirements = [req for req in textgen_requirements if 'bdashore3/flash-attention' not in req]
textgen_requirements = [req for req in textgen_requirements if 'jllllll/flash-attention' not in req]
with open('temp_requirements.txt', 'w') as file:
file.write('\n'.join(textgen_requirements))