Sumit6307 opened a new pull request, #1152:
URL: https://github.com/apache/mahout/pull/1152

   This PR aligns with the 2026 Q1 Roadmap for QDP (Quantum Data Plane) 
regarding "Integration & Release" and "Input Format Support."
   
   Changes
   
   - New feature
   - Refactoring
   - Test
   
   
   Why
   Currently, the Quantum Data Plane (QDP) has tight integration with PyTorch 
but lacks native support for the Jax ecosystem, which is a leading framework 
for quantum machine learning research. Furthermore, the `encode` method was 
specialized for specific frameworks, making it harder to support other 
DLPack-compliant libraries like Cupy. This change enables Mahout to act as a 
truly universal Quantum Data Plane.
   
   How
   - Jax Integration: Created a new `jax.rs` module that detects Jax arrays and 
handles asynchronous synchronization via `block_until_ready()` before data 
extraction.
   
   - Generic DLPack Fallback: Updated the unified `encode` method in 
`engine.rs` to support any Python object that implements the standard DLPack 
protocol (`__dlpack__` and `__dlpack_device__`).
   
   - Framework Interoperability: Ensured that the logic is "non-invasive"—the 
QDP only attempts to synchronize Jax/DLPack objects when they are passed as 
input, maintaining zero overhead for other paths.
   
   - Improved Validation: Added robust error handling during the DLPack 
extraction process to ensure device and memory alignment consistency.
   
   - Tests: Added a new test suite in `testing/qdp/test_jax.py` specifically 
for Jax-to-QDP-to-PyTorch workflows.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to