Initial dpctl tensor removal #2779
Conversation
|
/intelci: run |
Codecov Report✅ All modified and coverable lines are covered by tests.
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 5 files with indirect coverage changes 🚀 New features to boost your workflow:
|
ethanglaser
requested review from
Vika-F,
ahuber21,
david-cortes-intel,
icfaust,
napetrov,
syakov-intel and
yuejiaointel
as code owners
| print( | ||
| "Accuracy for entire rank {} (256 classes): {}\n".format( | ||
| rank, accuracy_score(y_test, dpt.to_numpy(y_predict)) | ||
| rank, accuracy_score(y_test, dpnp.asnumpy(y_predict)) |
There was a problem hiding this comment.
accuracy_score supports array API, so it should be possible to pass dpnp arrays to it:
https://scikit-learn.org/stable/modules/array_api.html#metrics
| # ============================================================================== | ||
|
|
||
| # sklearnex IncrementalPCA example for GPU offloading with DPCtl usm ndarray: | ||
| # sklearnex IncrementalPCA example for GPU offloading with DPNP ndarray: |
There was a problem hiding this comment.
Does it make sense to have examples that offload to GPU through DPNP arrays if there's already support for array API that can avoid transferring data back and forth with a one-liner change?
| "MSE for entire rank {}: {}\n".format( | ||
| rank, | ||
| mean_squared_error(y_test, dpt.to_numpy(y_predict)), | ||
| mean_squared_error(y_test, dpnp.asnumpy(y_predict)), |
There was a problem hiding this comment.
Also supports array API on the sklearn side.
| dpnp_X_train = dpnp.asarray(X_train, usm_type="device", sycl_queue=q) | ||
| dpnp_y_train = dpnp.asarray(y_train, usm_type="device", sycl_queue=q) | ||
| dpnp_X_test = dpnp.asarray(X_test, usm_type="device", sycl_queue=q) | ||
| dpnp_y_test = dpnp.asarray(y_test, usm_type="device", sycl_queue=q) |
There was a problem hiding this comment.
In the previous examples, y_test was left as a numpy array, and then y_predict converted to numpy to match with it. Here is is using y_test as dpnp and y_predict as numpy.
2 participants
Description
Replacement for #2764 but without modifications to internal dpctl tensor handling. Removes to-be-deprecated dpctl tensor usage in examples and tests, replacing with dpnp array use where appropriate.
Checklist:
Completeness and readability
Testing