scaleapi / scaleapi-python-client Goto Github PK

as stated in the requests API doc:
'''
Nearly all production code should use this parameter in nearly all requests. Failure to do so can cause your program to hang indefinitely:
'''
In the Scale Client class:

No timeout value is set or exposed, this may leave the caller hang in forever for a timeout

When I use scaleAI's deployment endpoint, if the generated text doesnt get sent back within ~30s, I get an error about a 504 error.

If the response comes back quickly, I dont.

This makes it very difficult to use for long text generation from GPTx's. Setting timeout in my request to 5mins does not change it, so I think its on Scale's end

I forgot to add the metadata of the filename while annotating my images, however, in the API it looks like there's no way to download the original image. All we get is the metadata of this form (Shown below), is there a way to convert it to a downloadable URL?

scaledata://61deda042221460099995568/e8cc0ceb-991e-45e8-9a5b-88827143da40

When downloading annotated tasks, I want to fetch only ones with accepted customer review status. Currently, this property will be added to the response payload only when the image is reviewed. However Task.py throws an exception because of how __getattr__ method is implemented. Ideally, it should return "nonspecified" or just empty string "" indicating there is not customer review yet.

Aside, I suggest avoiding this confusion at all by doing filtering on server side by updating list all tasks api to provide a filter based on customer_review_status

https://pypi.org/project/enum34/ is not needed on python 3.4 or greater, and installing it on Python 3.7 can cause all kinds of issues such as python-poetry/poetry#1122

This can be fixed by using environment markers using https://www.python.org/dev/peps/pep-0508/

I wanted to check the correct format for the hypothesis for a task. I checked the API documentation and couldn't find the right reference for it. Should it be an image that has pixel values as labels for the task? or should it be something else?

The current supported optional parameters for the function get_batches() are project_name, batch_status, created_after and created_before. However the API documentation shows more optional query params like audited_after, limit, offset, etc.
Will the method be supporting these parameters in the future?

Hi,

I noticed that the TaskReviewStatus enum (defined here) is missing the state skipped.

However, this may happen during auditing, namely tasks can be skipped and the status will be skipped in the resulting json.