@ancient-wolf-19325 is the following code supposed to work? How do I register a local task programmatically?

import pandas as pd
from sklearn.datasets import load_wine

from flytekit import task, workflow
from flytekit.configuration import ImageConfig, Config, SerializationSettings
from flytekit.types.pickle import FlytePickle
from flytekit.remote import FlyteRemote


# 🧱 @task decorators define the building blocks of your pipeline
@task
def get_data_new(x:int) -> pd.DataFrame:
    """Get the wine dataset."""
    print('hi')
    return load_wine(as_frame=True).frame



if __name__ == "__main__":
    # FlyteRemote object is the main entrypoint to API
    remote = FlyteRemote(
        config=Config.auto(),
        default_project="flytesnacks",
        default_domain="development",
    )

    remote.register_task(
        get_data_new, 
        serialization_settings=SerializationSettings(image_config=ImageConfig.auto(img_name="<http://ghcr.io/flyteorg/flytekit:py3.9-1.2.3|ghcr.io/flyteorg/flytekit:py3.9-1.2.3>")),
        version="v2")

    task = remote.fetch_task(name="get_data_new", version="v2")

    remote.execute(
        task,
        inputs={'x':0}
        )

The provided code snippet contains a few issues that need to be addressed for it to work correctly. Specifically, the

@task

decorator should be used correctly, and the

remote.register_task

method should be called with the appropriate parameters. Additionally, the

FlyteRemote

configuration and task execution should be properly set up. Here's a corrected version of the code:

import pandas as pd
from sklearn.datasets import load_wine

from flytekit import task, workflow
from flytekit.configuration import ImageConfig, Config, SerializationSettings
from flytekit.types.pickle import FlytePickle
from flytekit.remote import FlyteRemote


# 🧱 @task decorators define the building blocks of your pipeline
@task
def get_data_new(x: int) -> pd.DataFrame:
    """Get the wine dataset."""
    print('hi')
    return load_wine(as_frame=True).frame


if __name__ == "__main__":
    # FlyteRemote object is the main entrypoint to API
    remote = FlyteRemote(
        config=Config.auto(),
        default_project="flytesnacks",
        default_domain="development",
    )

    # Register the task
    remote.register_task(
        entity=get_data_new,
        serialization_settings=SerializationSettings(image_config=ImageConfig.auto(img_name="ghcr.io/flyteorg/flytekit:py3.9-1.2.3")),
        version="v2"
    )

    # Fetch the registered task
    task = remote.fetch_task(name="get_data_new", version="v2")

    # Execute the task
    remote.execute(
        task,
        inputs={'x': 0},
        wait=True
    )

Key Changes: 1. Corrected Syntax: Changed

->

to

->

in the

@task

decorator. 2. Image Configuration: Corrected the

img_name

parameter in

ImageConfig.auto

. 3. Register Task: Used the

entity

parameter to register the task. 4. Execution: Added

wait=True

to the

remote.execute

method to wait for the task execution to complete. For more details, refer to the FlyteRemote documentation. Would you like more details on any specific part of the process? Sources: - FlyteRemote documentation - FlyteRemote API Reference 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.

@miniature-traffic-86643 can you run it using —- pyflyte run filename get_data_new —x 10

yes, that works

Ohh you can, did you try remote.execute?

yes, I also tried execute directly, without register. Still no luck.