Notebook environment

Introduction

The notebook is designed to use all-cpu / all-gpu depending on the selected instant type. These images include typical machine learning and deep learning libraries / frameworks, so you can use these libraries / frameworks on your notebook.

See all-cpu / all-gpu for the library / framework installed in each image.

Use SDK

ABEJA SDK is installed in the notebook by default.

Tokens that have the authority to access resources of the same organization are also set in environment variables. You can be used as follows.

from abeja.datalake import Client

# Channel ID in same organization
channel_id = '1234567890123'

# Client initialization
# Initialize using the credentials set in the notebook
client = Client()
# Channel information can be acquired with the set authentication information
channel = client.get_channel(channel_id)

Use CLI

The notebook has ABEJA CLI installed by default.

It can be used from the terminal screen of the notebook or Command execution in IPython.

Please refer to ABEJA PLATFORM CLI for how to use CLI.

Data persistence

The data existing under the / mnt directive remains. However, any data saved outside the / mnt directory will be deleted as well.

The notebook’s home directory is set to / mnt / notebooks / {notebook_id}. The files created with the notebook will remain even if stopped. Also, under the / mnt directory, shared directories among the same organization, It is possible to reuse files saved in a notebook from other notebooks and training jobs.