alisha-1000 commented on issue #719: URL: https://github.com/apache/mahout/issues/719#issuecomment-3939383164
Hi @rich7420, I’m interested in contributing to TensorFlow tensor input support for QDP. From the issue breakdown, it seems the work is split into: #728: Eager tensor input support with CPU/GPU detection #729: GPU support via DLPack and device placement #730: TF-specific optimizations and documentation I’d be happy to start with #728 (eager tensor input support) as a first step. My understanding is: Detect tf.Tensor type Handle eager tensors directly Detect device placement (/CPU:0, /GPU:0) Convert appropriately to QDP’s internal format (likely via NumPy or DLPack depending on device) Before starting, could you confirm: Whether CPU-only TF tensor support is preferred as Phase 1? Whether DLPack is the required pathway for GPU tensors? Any constraints regarding zero-copy transfers? I’d like to align with the roadmap (#752 / #799) before beginning implementation. -- 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]
