ci: 👷 release,test-release,dep actions and pyproject.toml added #43
Conversation
Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
|
|
Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
onuralpszr
force-pushed
the
ci/pypa/dep
branch
from
805166e to
3e1893f
Compare
|
Current test results py39: OK (66.17=setup[32.69]+cmd[33.48] seconds)
py310: OK (57.87=setup[26.58]+cmd[31.29] seconds)
py311: OK (55.35=setup[25.66]+cmd[29.69] seconds)
py312: OK (61.36=setup[30.26]+cmd[31.10] seconds)
congratulations :) (240.78 seconds)It also confirms installation work properly |
Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
onuralpszr
force-pushed
the
ci/pypa/dep
branch
2 times, most recently
from
e38869b to
4110122
Compare
Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
onuralpszr
force-pushed
the
ci/pypa/dep
branch
from
4110122 to
734980d
Compare
|
Hi @SkalskiP 👋, I've completed most of the actions for Maestro and replaced the setup.py and requirements.txt files with a pyproject.toml, leveraging setuptools. Additionally, I wrote actions for both macOS and Ubuntu to ensure that all installations and tests pass. I manually added a FlashAttention wheel for a clean installation in test CI but PyPI version of flash-attention only provides a tarball that requires compilation. However, this complicates a one-liner installation (pip install maestro) since users would need to either set up the proper environment or install flash-attention manually alongside Torch.
|
Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
onuralpszr
force-pushed
the
ci/pypa/dep
branch
from
5183d76 to
d1919fc
Compare
- 📝 Added a new CHANGELOG.md file to document changes. - 📝 Updated README.md with new information or corrections. - 🔧 Modified pyproject.toml to update project urls and exclude corrections - 🔧 Package keywords added to pyproject.toml Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
Wheel Package Content listadding 'maestro/init.py' |
| pip install torch | ||
| if [[ "${{ matrix.os }}" == "ubuntu-latest" ]]; then | ||
| if [[ "${{ matrix.python-version }}" == "3.10" ]]; then | ||
| pip install https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiTRUE-cp310-cp310-linux_x86_64.whl |
There was a problem hiding this comment.
will this break when torch is bumped to next version given flashattn is fixed to 2.4?
There was a problem hiding this comment.
Yes, we need to update here, or I can lock pytorch in here and we can update when we wanted to as well. But source installation was breaking in action and I checked documentation of flash-attention it was asking other packages and cuda-toolkit to be present plus they also have certain version lock as you can see in pre-made as wheels
Description