Toby Roseman

@dhgokul - There are significant performance differences between 6.2 and 6.2.2 for object detection, if you are using Linux. These are also present, if you are training on CPU or...

TuriCreate is a Python package which wraps C++ code. The line number you share is simply the `group_by` entry point into C++. So knowing that it fails there isn't very...

@hoytak - will this change impact our users? Is there functionality where the same data and the same seed will now produce different results? If so, what functionality is affected?

I believe that site has been deactivated. Is there a reason you still need to use the academic version of GraphLab Create rather than the open source version of TuriCreate?

@zahragis - it looks like you are using Windows as your operating system. TuriCreate does not natively support Windows, see #533. If you are using Windows 10, you can use...

With the release of TuriCreate 6.0 all deep learning toolkit are now using TensorFlow. Also TuriCreate has as always been able to run on Linux. @Zumbalamambo - What are you...

This is related to #1141. @Zumbalamambo - you could try using the ONNX approach mentioned in that issue.

> Answering your question posted in #3388 , I want to export any of the recommender models. Is it possible ?, how I could do it ? Thanks for the...

> Basically I'd like to create a flutter app with real time image detection and I was thinking of using your framework, after training a mobilenet, is it possible to...

TuriCreate does not have native M1 support, but it should run fine on an M1 machine if you use Rosetta. See my [instructions](https://github.com/apple/turicreate/issues/3429#issuecomment-962288689) in another GitHub issue to use TuriCreate...