cleanlab / cleanlab-studio Goto Github PK

Client interface for all things Cleanlab Studio

License: MIT License

Python 99.29% Makefile 0.71%

annotations automl computer-vision data-centric-ai data-labeling data-profiling data-quality data-science data-validation machine-learning

cleanlab-studio's People

Contributors

Stargazers

Watchers

Forkers

anishathalye jwmueller elisno tqdo

cleanlab-studio's Issues

Add timeout option to get_confidence_score call for TLM with a reasonable default time

Feature embeddings are not the same length as the dataset

the feature embeddings are not the same length as metadata.csv when we add a zipped image dataset, could be indexing/something is dropped over the pipeline.

Add Public Documentation for Image Support

Add Instructions to use Cleanlab-Studio CLI to upload image datasets and how to run cleanlab against image datasets.

Fix type checking

I think this type annotation should just be str, not Optional[str]: https://github.com/cleanlab/cleanlab-studio/pull/79/files#diff-267c264a8f2bb5a51ed4f82a9fcdf904079440739bed7c67e74299c879bdcc7eR488

Cleanlab CLI upload dead link

Upon uploading a dataset, the CLI prints:

No issues were encountered when uploading your dataset. Nice!
Upload completed. View your uploaded dataset at https://app.cleanlab.ai/datasets

This results in a 404 error. Just point to http://app.cleanlab.ai instead?

Handle $, % in integer strings

Similar to #32, it would be helpful to include functions extract integers from dollar and percentage values.

API to delete dataset

Can we have an API to delete dataset?

Release new README to pypi

At some point, need to re-release so the readme on https://pypi.org/project/cleanlab-cli/ gets updated with the current readme on github

Interchange (<dataset>, <cleanset_id>)

As per the API docs, in the apply_corrections method, the <cleanset_id> is the first argument and second.
The following line in README has switched them -

cleanlab-studio/README.md

Line 45 in c25cce6

    
           corrected_dataset = studio.apply_corrections(<your dataset>, <your cleanset id>)

Unable to get Cleanlab API

I am trying to do the Cleanlab Python API but can't get the API key.
It shows something like this even though I have signed up for the account by following the process given in the documentation.

Please help me fix this issue, thanks a ton in advance.

Error messages from missing editors

The CLI currently shows:

$ cleanlab dataset upload ...
...
Would you like to open your issues file for viewing? [y/n]: y
/bin/sh: atom: command not found
/bin/sh: subl: command not found
/bin/sh: code: command not found

At the very least, we should figure out how to suppress this output.

Why do we supply recommended editors anyways? Can we not just use click.edit(filename=...) and have it rely on $EDITOR to open the right editor?

Add social share preview card to app.cleanlab.ai

Our social share preview card appears to be gone.

https://www.linkedin.com/post-inspector/inspect/https:%2F%2Fapp.cleanlab.ai%2F

Here is an updated one:

Figma here

Project Endpoint

Could we have an endpoint that lists Project IDs, cleanset IDs, dataset info, and the time of run?
It'll be helpful to delete multiple projects with an endpoint