Cyprien

[FEATURE] Handling extras

7

Hi, Would it please be possible to handle extras (like in `pip install pettingzoo[atari]`) when generating the recipe? Ideally, doing `grayskull pypi pettingzoo[atari]` would fetch the extras and add them...

Update python-usages_examples.md

4

Hello, It seemed to me that the batch training example had a typo. The baseline for the second batch should be the predictions of the first model for the second...

feat: manylinux_2_34 based on AlmaLinux 9

1

Hi, I imagine this will require more discussions but I thought I'd at least open an MR proposing a `manylinux_2_34` compliant image based on AlmaLinux 9. I need it to...

Error Repairing Wheel to manylinux2014_x86_64 ABI Due to Too-Recent Versioned Symbols

2

When attempting to repair the wheel for `triton_nightly` to the `manylinux2014_x86_64` ABI, the auditwheel tool fails with the following error: ``` auditwheel: error: cannot repair "/tmp/cibuildwheel/built_wheel/triton_nightly-3.0.0.post20240212030344-cp38-cp38-linux_x86_64.whl" to "manylinux2014_x86_64" ABI because...

Release v0.0.27 to conda forge

1

Hi, It looks like the latest version (v0.0.27) was not released to [conda-forge](https://anaconda.org/conda-forge/sseclient). Could it please be possible to release it? Thanks, Cyprien

DynamicLossScale.min_loss_scale isn't passed in the DynamicLossScale.tree_flatten outputs which triggered the default factory when calling DynamicLossScale.tree_unflatten. Potentially changing the dtype while doing so (and of course, not keeping the original value).

Build from source on Apple M1

4

Hi, I'm trying to build launchpad from sources on a laptop with the M1 chip. Is it possible? I've tried using the Dockerfile provided - unsuccessfully. Curious to known if...

Hi, It looked like the for-loop used for installing tensorflow and dependencies for each requested python versions didn't point to a valid command. I've replaced each `$python ...` with `python${python}...

TypingError When Accessing Attributes of Nested JitClasses

4

## Reporting a bug - [x] I have tried using the latest released version of Numba (0.60.0). - [x] I have included a self-contained code sample to reproduce the problem....

fix: removing .encode() in call to register_custom_call_target

1

The function `envpool.python.xla_template._make_xla_function` wrongfully passes `bytes` instead of `str` for the `name` parameter when calling `xla_client.register_custom_call_target`. This leads to errors when calling `envs.xla()`. Removing `.encode()` fixes the issue.