altanh commented on pull request #8307: URL: https://github.com/apache/tvm/pull/8307#issuecomment-867054296
> This looks interesting. Do we have a RFC or a tracking issue that indicates where the Training support of TVM is going ? There's no RFC up at the moment, but we've had some discussion threads (see https://discuss.tvm.apache.org/t/two-missing-pieces-for-training/10037 for example). I'm currently in the process of upstreaming general improvements (gradients, AD bug fixes / improvements, new training ops, other supporting changes like contrib library support) and longer term hoping to open source a proof-of-concept TVM training framework around Q3 (developed at OctoML). Would love to hear your thoughts and perhaps it would be worth opening a long term tracking/discussion thread for training related topics! cc @tqchen -- 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. For queries about this service, please contact Infrastructure at: [email protected]
