Contents x
      Start Jobs with LandingLens CLI
      • 21 Dec 2022
      • 1 Minute to read
      • Print
      • Dark
        Light
      • PDF
      Contents

      Start Jobs with LandingLens CLI

      • Updated on 21 Dec 2022
      • 1 Minute to read
      • Print
      • Dark
        Light
      • PDF
      Article summary

      Overview

      You can start jobs with LandingLens CLI or Python SDK. Launching jobs via LandingLens CLI has these advantages:

      • Programmatically initiate jobs
      • Fine-tune Model hyperparameters
      • Apply custom transformations in pre- and post-processing
      Note:
      The maximum number of runnings jobs on the Project-level is 10. If you have 10 jobs running, then launching a new job will fail.

      Initiate Source Codes

      In a new local folder, run the following:

      llens model init object-detection  # or segmentation

      This will create the following files and folders that contain source codes to launch a LandingLens CLI job:

      .
      ├── README.md
      ├── custom
      │   ├── __init__.py
      │   └── object_detection_to_classification.py
      ├── test
      │   └── test_od.py
      ├── train.yaml
      └── transforms.yaml

      Configure Jobs

      To configure your job:

      1. Configure the train specifications.
      2. Create and apply custom transformations.

      Launch Train-Evaluation Jobs

      Command

      llens model train <FOLDER> --dataset-id <DATASET_ID> --job-name <JOB_NAME> (--job-description <JOB_DESCRIPTION>)

      Parameters

      ParameterDescription
      FOLDERThe directory that contains train.yaml and transforms.yaml. If you are in the same folder of these files, you can use “.” to denote the current folder.
      DATASET_IDDataset ID number. You can find the ID of an exported dataset from LandingLens Platform under Data > Actions (horizontal ellipses) > Exported Dataset.
      JOB_NAMEThe name of the job.
      (Optional) JOB_DESCRIPTIONThe description of the job.

      Launch Evaluation-Only Jobs

      Command

      llens model eval <FOLDER> --dataset-id <DATASET_ID> --train-job-id <TRAIN_JOB_ID> --job-name <JOB_NAME> \
          (--job-description <JOB_DESCRIPTION>)

      Parameters

      ParameterDescription
      FOLDERThe directory that contains train.yaml and transforms.yaml. If you are in the same folder of these files, you can use “.” to denote the current folder.
      DATASET_IDDataset ID number. You can find the ID of an exported dataset from LandingLens Platform under Data > Actions (horizontal ellipses) > Exported Dataset.
      TRAIN_JOB_IDThe ID of the train job. You can retrieve this ID from the URL of the Model Detail page. Example: https://platform.landingai.io/app/{ORG_ID/pr/{PROJ_ID}/models/view/{TRAIN_JOB_ID}.
      JOB_NAMEThe name of the job.
      (Optional) JOB_DESCRIPTIONThe description of the job.
      Was this article helpful?
      What's Next