mlops

Describe the bug
I am trying to label Hebrew text (RTL language). When labels are attached to the text, the words of the text are mixed and not shown in their original order.

To Reproduce
Steps to reproduce the behavior:

1. Create a project with attached dataset.json dataset.txt
2. Choose NER template
3. Start

Description

We're running usability tests and would love for you to record walking through our tutorials. The idea for this ticket is that you do a screen capture walking through one of more of the following examples:

• Hello World! (15 minutes)
• [Iris Dataset](https://kedro.readthedocs.io/en/stable/02_get_started

Describe the bug
data docs columns shrink to 1 character width with long query

To Reproduce
Steps to reproduce the behavior:

1. make a batch from a long query string
2. run validation
3. render result to data docs
4. See screenshot
   <img width="1525" alt="Data_documentation_compiled_by_Great_Expectations" src="https://user-images.githubusercontent.com/928247/103230647-30eca500-4

We currently have read and write capabilities but do not support deleting. We could add a few calls like delete delete_all and some recursive way of deleting.

🚨🚨 Feature Request

• A new implementation (Improvement, Extension)

Is your feature request related to a problem?

Currently, if a user tries to access an index that is larger than the dataset length or tensor length, an internal error is thrown which is not easy to understand.

Description of the possible solution

We can catch the error and throw a more descriptive e

Is your feature request related to a problem? Please describe.
The current Feast online store for GCP implementation requires Firestore in Datastore mode. Firestore can only be in one mode at a time per GCP account. You cannot use native mode for some applications and Datastore mode for others within the same account. Adding a feature store to an existing GCP account that uses native mode wou

The documentation for creating complex inference graphs should include what components can be connected to what, and how the overall graph should behave.

For example, all inference graphs need to end with a single node, whether its a combiner combining multiple inputs, or a transformer, or a model--the graph can't split and then never rejoin, etc.

Use case

Polyaxon tracking can be used to automatically log all information generated by Ludwig.

We should probably think about adding support for Polyaxon's tracking module in Ludwig contrib.

I have uploaded the metrics_visualization pipeline in the Kubeflow Console and when I invoke a run I am able to see the visualizations yet they are not seen in the run output

![image](https://user-images.githubusercontent.com/59866888/153585857-cda2921e-48b9-4528-b568-e8866fffb993.

🚀 The feature

As far as I know, there are no examples or documentation on serving Speech2Text models from Huggingface, such as Wav2Vec2. How could I enable serving with Wav2Vec2 Huggingface pre-trained checkpoints ?

Motivation, pitch

I'm working on Huggingface Wav2Vec2 models' deployment, and would like serving on these models made possible. Thank you !

Alternatives

No response

🚀 Feature

There is no out-of-box support for .aim repository backups. Users have to manually create an archive from repo directory and upload to cloud storage buckets.

Motivation

A single-command backup of aim repos. Pre-defined clear naming/versioning across all repos.

Pitch

Add a CLI commands to backup and restore .aim repo in/from cloud storage. Support S3 and GCS. Backup

I added a small tutorial (triggered with python -m ploomber.onboard) but it doesn't have tests yet

TL;DR

With all the work behind us in pyflyte package, run, and register, along with the legacy serialize, the pyflyte code could probably use some cleanup.

Details

This ticket includes but shouldn't be construed as to be limited to the following:

• Use the same module loading code between the run and register commands. Since run has to first register, and since it also

Describe changes

I fixed issue : Log a warning instead of raising an AssertionError

else:
    # TODO [ENG-895]: Refactor this to raise a warning instead.
    raise AssertionError(
        "Cannot set magic flag in non-notebook environments."
    )

to

else:
    cli_utils.warning(
        "Cannot set magic flag in non-notebook environments."
    )

Describe the solution you'd like
Currently, there are some RST files that contain code examples within a .. code-block. This code blocks may break when we implement new features to the package and break the API.
We need some mechanism that validates the code blocks in the RST docs are valid, and fail the PR check if something is broken.

We can use docutils or some other alternative for

add the workflow to release the installing package of macos automatically when a new tag created

Currrently , the CLI outputs "No such file or directory" but should be more verbose and explain which file could not be found.

No such file or directory (os error 2)