maheshambule opened a new pull request #5052: Relay to ONNX converter URL: https://github.com/apache/incubator-tvm/pull/5052 This PR adds support to convert In-memory Relay Module to ONNX protobuf. Currently, only a subset of operators is supported. The conversion can be done on optimized Relay modules as well except for fused op passes. Operators supported: - [x] reshape - [x] conv2d - [x] add - [x] relu - [x] transpose - [x] dense - [x] max_pool2d - [x] batch_flatten - [x] multiply - [x] bias_add - [x] batch_norm - [x] global_avg_pool2d - [x] concatenate - [x] dropout - [x] avg_pool2d - [x] divide - [x] mean - [x] pad - [x] softmax - [x] squeeze - [x] strided_slice - [x] greater - [x] less - [x] equal - [x] zeros_like - [x] ones_like - [x] subtract Models tested: - [x] ONNX zoo resent and squeezenet - [x] TF-Slim Resnet - [x] MXNet Resnet TO-DO: - Increase the operator coverage to add support for all the Relay operators - Add support for optimization pass - AlterOpLayout - Add support for Relay constructs such as functions, ADTs etc - Add support for different versions of ONNX opsets for different operators
---------------------------------------------------------------- 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] With regards, Apache Git Services
