mdespriee commented on issue #11926: segfault in native code while trying to use CustomOp URL: https://github.com/apache/incubator-mxnet/issues/11926#issuecomment-409932193 @nswamy Sure. We have a quite big predictive pipeline, working on CRM data (we're in the marketing intelligence business). It's mainly based on an unsupervised phase (clustering, collaborative...) followed by a logistic reg. This pipeline has been built for years, using quite low level operations (linear algebra, in python). We'd like to evolve our technologies and approach. Not surprisingly the supervised part is a DAG of operations with some similarity to a DAG of Symbol. We also would like to experiment in a near future the integration of an LSTM to process some features. And I'm interested in building bridges with the scala/jvm world. MXNet looks like a good candidate for the task, doesn't it ? So that's the reason of my experiments. The use of constants is one of our requirements (we have many of them everywhere in our current DAG), and also we're heavy users of sparse computations. That's why I'm also very interested in your roadmap regarding this topic.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services