AccernXYME is a library for easily accessing XYME via python.


Keywords
XYME AI machine learning client
License
MIT
Install
pip install accern-xyme==0.0.17

Documentation

Accern-XYME

accern_xyme is a python library for accessing XYME functionality.

CircleCI

Usage

You can install accern_xyme with pip:

pip install --user accern-xyme

Import it in python via:

from accern_xyme import create_xyme_client

client = accern_xyme.create_xyme_client(
    "https://xyme.accern.com/", "<USERNAME>", "<PASSWORD>")
print(client.get_user_info())

<USERNAME> and <PASSWORD> are the login credentials for XYME. The values can also be set to None in which case the values must be set in the environment variables ACCERN_USER and ACCERN_PASSWORD. A login token can also be provided.

You will need python3.6 or later.

Exploring Workspaces

The workspaces of the user can be retrieved via:

for (workspace, count) in client.get_workspaces().items():
    print(f"{workspace} contains {count} jobs"

And jobs in a given workspace can be retrieved via:

for job in client.get_jobs(workspace):
    print(f"{job.get_job_id()}: {job.get_name()} - {job.get_status()}"

Or directly by Job ID:

job = client.get_job_id("username_example_com/job_id")

Starting Jobs

A new job can be started via:

# creating the job
job = client.create_job(schema=schema_obj, name="my job")

with job.update_schema() as cur:
    # updating the schema
    cur["M"]["params"]["hidden_layer_sizes"] = [100, 100, 100]

# starting the job
job.start()

import time
time.sleep(30)

print(job.get_status())

Computing Predictions

Predictions can be obtained for a finished or running job:

# predict_proba is also available
predictions, stdout = job.predict(df)
print(stdout)

print("prediction of first row: ", predictions.iloc[0])