[OpenVINO] Fix for SparseCategoricalCrossentropy: cast target to int before one_hot

[goyaladitya05]

OpenVINO's one_hot op requires integer indices, but target in sparse_categorical_crossentropy was passed without casting, causing failures when float labels were provided.

This PR adds casting target to int64 unconditionally at the top of sparse_categorical_crossentropy, similar to how other backends do it.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.28%. Comparing base (6c5956d) to head (38251a4).

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #22607   +/-   ##
=======================================
  Coverage   83.28%   83.28%           
=======================================
  Files         596      596           
  Lines       68102    68103    +1     
  Branches    10610    10610           
=======================================
+ Hits        56720    56721    +1     
  Misses       8636     8636           
  Partials     2746     2746           

Flag	Coverage Δ
keras	83.09% <100.00%> (+<0.01%)	⬆️
keras-jax	59.65% <0.00%> (-0.01%)	⬇️
keras-numpy	55.33% <0.00%> (-0.01%)	⬇️
keras-openvino	53.41% <100.00%> (+0.02%)	⬆️
keras-tensorflow	61.03% <0.00%> (-0.01%)	⬇️
keras-torch	59.84% <0.00%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[gemini-code-assist]

Code Review

This pull request enables several SparseCategoricalCrossentropy tests for the OpenVINO backend by introducing an unconditional cast of the target tensor to i64 within the sparse_categorical_crossentropy implementation. While this change allows the tests to pass, the feedback indicates that hardcoding this cast within the loss function is an internal implementation detail that violates Keras API design guidelines. It is recommended to handle such type conversions through the backend's type promotion logic instead of within the specific function implementation.

[hertschuh]

Thanks for the fix!