Trying to install

[MatthewGCampbell]

Which way would you recommend installing turicreate preferably the easiest way. Tried using Jupiter Notebook like in the WWDC 2018 video, but I was having trouble importing turicreate into the environment. It always said that my python is too new and that I need v.3.6 bu ti have 3.7. does anyone know how to downgrade python using anaconda?

[MatthewGCampbell]

now I'm getting this issue

[TobyRoseman]

Turi Create doesn't yet support Python 3.7. That issue is being tracked in #788.

Based on the paths in your stack trace, it looks like you've figured out how to create a Python 3.6 environment using anaconda.

I'm not sure what's happening with your second issue. Please share the entire stack trace. From just the part you've shared I can't tell what is happening.

[MatthewGCampbell]

there's my error code

[MatthewGCampbell]

https://onedrive.live.com/view.aspx?resid=5CA41B473B8128E1!4713&ithint=file%2cdocx&authkey=!AEgUJnr9vdjyZr8

[znation]

@MatthewGCampbell It looks like it's failing to parse some weird characters in your CSV file: ,[1;31m^[0m. Is it possible it's a corrupted download, bad copy/paste, etc.? Are you able to open this CSV file and have it parsed properly in any spreadsheet application, like Excel, Numbers, etc.?

[MatthewGCampbell]

@znation did you recieve my email? and I got it to work on my MacBook Air with the same file just not on my MacBook Pro ill send you a Excel screenshot

[MatthewGCampbell]

[znation]

Hi @MatthewGCampbell From the screenshot, the CSV file itself looks fine. Could you share the CSV file with us (either in this thread, or over email) and we can help diagnose the issue?

[MatthewGCampbell]

Sorry I’m fixing one issue at a time right after I took the screenshot I saved it through excel and now it seems to work fine! But my next problem it that I created the model but can’t seem to find it on my computer, yeah I’d be happy to email it’s more simple than using github my email is: [email protected]

[MatthewGCampbell]

Is their like a default file name or place that it could be located

[znation]

@MatthewGCampbell Once you have a model in Python, you would assign it to a variable like:

model = tc.classifier.create(...)

Then, once it's in a model variable, you can save it to disk or export it for use with CoreML. You must explicitly determine the path in both cases. If you don't call save or export_coreml, your model will not be saved to disk at all.

model.save('path/to/turicreate_model') # saves in Turi Create model format
model.export_coreml('path/to/model.mlmodel') # exports in Core ML model format

Hope that helps!

[MatthewGCampbell]

Thanks I don’t remember what code I ran and don’t have access to my laptop cause I’m n my car but I ran something like model = tc.classifier.create(...) but I replaced the ... with data so after I run that command it gave me this https://share.icloud.com/photos/0fdoKoY8K3armb5jsRwqhK66A and after I run that command would I have to run the model.save command and the model export one too?

[MatthewGCampbell]

Which email is best for you could you send me an email from that address? since I usually don’t check github and the easiest way for me is email and I respond quicker to my email.

[znation]

@MatthewGCampbell It's better to keep issues on GitHub, as that way the rest of the community can keep aware of what issues we're having, weigh in on possible fixes and workarounds, and provide more information when they are also able to reproduce the issue. If you're having trouble getting notifications from GitHub over email, see this article on how to configure notifications.

I think the answer to your latest question is, yes, you would need to call save and/or export_coreml after the create command. Depending on how you want to preserve the model artifact to disk and where you intend to use it, one or the other or both could make sense. save will save a model that is loadable in Turi Create (but not Core ML directly), and export_coreml will save a model that is loadable in Core ML (but not Turi Create).

[MatthewGCampbell]

So then could I run both? Makes since to keep it in github tho.

[znation]

@MatthewGCampbell Sure, you can run both. They will produce different file formats. The save method produces a Turi Create-compatible model in a directory structure. The export_coreml method produces a Core ML model in the .mlmodel file format.

[MatthewGCampbell]

here is the command I ran for the model

[MatthewGCampbell]

im a high school student and I'm going to a swift developing summer camp and the directors were asking if next year you would mind being a guest speaker? that's why I needed your email if you can't then no problem

[MatthewGCampbell]

The model.save and the model.export_coreml commands don't seem to work not sure why? do the files already have to exist or is it creating a new file?

[TobyRoseman]

When you seen In [ * ]: next to the code input box that means it's still working. When it's done the * will be replaced by a number.

[MatthewGCampbell]

if I call it model 1 in the tc.object_detector.create command does it need to bet called model 1 for the rest of the code?

[MatthewGCampbell]

bet = be

[MatthewGCampbell]

Got it to work finally now can you just import the ml file

[MatthewGCampbell]

trying to run this stuff from your WWDC talk when running the SFrame command how do I create the file with the images that it predicts?

[MatthewGCampbell]

what kind of a file is the .sframe? how do you make it?

[TobyRoseman]

Please study the Introduction to SFrames and Object Detection section of the user guide.

Since the original issues have been resolved I'm going to close this issue. If you have additional questions after studying our user guide, open a new issue.

[MatthewGCampbell]

Thanks so much for your help