Add max_pool and average_pool for MLX #20814
Conversation
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
keras/src/backend/mlx/nn.py
Outdated
| raise ValueError( | ||
| f"To extract sliding windows the window shapes and strides must " | ||
| f"have the same number of spatial dimensions as the signal but " | ||
| f"the signal has {len(spatial_dims)} dims and the window shape " |
There was a problem hiding this comment.
Note: lines that don't have {}s in them don't need to start with f. This applies to all string lines in the PR.
In this error message, please format the argument information via name={value}, e.g.
"input_shape={x.shape}, window_shape={window_shape}, strides={window_strides}"
keras/src/backend/mlx/nn.py
Outdated
| raise ValueError( | ||
| "To extract sliding windows, the lengths of window_shape and " | ||
| "window_strides must be equal to the signal's spatial dimensions. " | ||
| f"However, the signal has spatial_dims={len(spatial_dims)} while " |
There was a problem hiding this comment.
For all args, rather than printing just len it's useful to print the full arg value
google-ml-butler
bot
added
kokoro:force-run
ready to pull
google-ml-butler
bot
removed
ready to pull
This PR implements support for max_pool and average_pool operations in the MLX backend.
To achieve this functionality the following functions were copied from mlx repository:
_non_overlapping_sliding_windows and _sliding_windows
The tests related to pooling in:
were passed