Last modified 2025-09-08

Review an AI/ML Training Run (Tutorial)

	Abbreviations Key
AI	artificial intelligence	JSON	Javascript Object Notation
HISE	Human Immune System Explorer	ML	machine learning
`hp`	hisepy	QC	quality control
IDE	integrated development environment	SDK	software development kit
int	integer	str	string

At a Glance

This document explains how to review and then approve or reject an AI/ML training job by calling the review_training_job_run() SDK method. The box below shows the review_training_job_run() signature, and the method parameters follow.

review_training_job_run()

hp.review_training_job_run(
    job_id: str,
    study_space_id: str,
    review_notes: str = None,
)

Parameter	Data type	Required or optional	Default (if any)	Description
`job_id`	`string`	required	NA	ID of the training job to be reviewed
`study_space_id`	`string`	required	NA	ID of the study space that contains the job to be reviewed
`review_notes`	`string`	optional	None	Notes for the review

Get Help

If you get stuck during a review_training_job_run() call, refer to the steps of this tutorial. To use the baked-in help in your IDE, try one of the following commands. Still not working? Click Support to file a ticket.

Python	Output
`help(hp.review_training_job_run)`	Function signature, list of parameters, class, and a brief description of the method in a compact plain-text format
`hp.review_training_job_run?`	Method signature, docstring (description), file location, and file type in more readable format
`hp.review_training_job_run??`	Signature, docstring, file path, a verbose set of metadata, and the source code for the method

Instructions

Import Libraries

To get started, set up your environment to interact with HISE programmatically and access all available SDK functions. For details, see Use Hise SDK Methods.

1. Navigate to HISE, and use your organizational email address to sign in.

2. Open an IDE. For instructions, see Create Your First HISE IDE (Tutorial).

3. Install the HISE SDK if you haven't already done so.

# Use pip to install the SDK
!pip install hisepy

4. To import hisepy, enter the following code into a new cell in your IDE.

# Import the Python SDK to enable programmatic access to HISE functions
import hisepy as hp

Run a Training Job

1. Run an AI/ML traning job, as described in Start an AI/ML Training Job (Tutorial).

Extract the Job ID

1. Extract the trainingJobId from start_training_run().

A. Navigate to the HISE AI/ML Hub.

B. Select Training Runs.

C. Find your run from Step 2, and click VIEW DETAILS.

D. Click the Process Info tab.

E. Copy the value of the trainingJobId field.

Call review_training_job_run()

1. In a new cell in your IDE, call hp.review_training_job_run() with all required and optional parameters.

A. Enter the trainingJobId you copied in Step 3 as the value of the job_id parameter.

B. Enter the study_space_id for the study space that contains the job to be reviewed. If you're not sure, use hp.get_study_spaces() to find your ID.

C. (Optional) Add a meaningful review note, especially when rejecting a run. This step is optional but highly recommended to promote transparency and traceability.

D. Run the cell.

hp.review_training_job_run(
    # The ID we extracted in Step 3
    job_id="f7b20726-2f94-4a31-9ba2-2b1760d2ca6d",
      # The study_space_id
    study_space_id="50744572-6f5f-4924-a5c7-367e0dee3b2a",   
    # Your review comments                    
    review_notes="Data preprocessing and parameter tuning are complete. The model 
passes all validation checks and is ready for integration. Training run approved."
)

Approve or Reject the Run

1. If your run is successful, you receive a message similar to the one below.

Training job output files downloaded to: ['input/condastageenv/2842984779/training_job_review/ray/f7b20726-2f94-4a31-9ba2-2b1760d2ca6d/1757350425/empty.txt']

Training job outputs have been downloaded to your IDE. Please review the outputs, and specify if you want to accept or reject the training job: f7b20726-2f94-4a31-9ba2-2b1760d2ca6d

Would you like to approve this job? (y/n) y

A. At the prompt, enter y to accept the run or n to reject the run. In this example, we choose y, and the following message appears.

{'Job': 'f7b20726-2f94-4a31-9ba2-2b1760d2ca6d',
 'approved': 'true',
 'message': 'Job Reviewed'}

B. If the call is not successful, a CreationFailure tag appears on the training run card. To identify the issue, check the IDE's error message, and see the Troubleshooting box below.

TROUBLESHOOTING

File not under review

Be sure to call start_training_run() before you execute review_training_job_run(). Only jobs that have been initialized and completed can be reviewed.

Invalid or missing job_id

This kind of error occurs if you supply a job_id that doesn't match an existing training job. Follow Step 3 to find the correct job_id generated by your start_training_run() call.

Review Environment and Training Run Details

1. In HISE, go to RESEARCH > AI/ML > Training Runs to watch your job start, view logs, and track progress.

A. Find your run, and click VIEW DETAILS. The page opens to the default PROPERTIES tab.

B. Inspect the properties details for your run, or choose among the remaining tabs. The function of each tab is summarized in the following table.

Training Run Details
Tab	Description	Function
PROPERTIES	Summarizes model, training run, and resource details, including entry point, status, and audit trail.	Confirm your experiment setup and troubleshoot issues related to training configuration.
DASHBOARD	Provides interactive subtabs (Overview, Jobs, Serve, Cluster, Actors) that let you drill down on the overall job status, individual jobs, serving configuration, cluster resources, and Ray actors.	Monitor live resource usage, track job progress, and diagnose distributed workflow performance.
WORKFLOW	Opens a direct link to the detailed workflow view in the Google Cloud Console.	Access full workflow logs, track execution steps, and find detailed troubleshooting information.
PACKAGES	Shows package information for the session if available; displays as unavailable if package info is not applicable to the current workflow.	Review the list of installed packages to ensure all dependencies are available.
PROCESS INFO	Shows complete process metadata in JSON format, including IDs, audit information, and configuration details for the run.	Export or check run metadata for support requests or debugging.
FILES INFO	Supplies the full JSON record, including metadata, source buckets, QC indicators, and file attributes, for each input file used in the run.	Examine file provenance, verify data quality, and ensure input files are correctly configured.

C. To review the environment configuration used in your training run, go to Research > AI/ML Dashboard > Environment Details. The page opens to the default PROPERTIES tab.

The table below summarizes all four available tabs.

Environment Details
Tab	Description	Function
PROPERTIES	Displays an overview of your environment, including its description, artifacts, Docker image, unique identifiers, and audit history.	Investigate environment setup, check reproducibility, and track when or how the environment was created or modified.
SYSTEM PACKAGES	Lists the core system-level packages installed for this environment, along with status and configuration metadata.	Verify foundational packages and their availability for your workflow.
BUILD INFO	Shows the build lifecycle of your environment, detailing key stages and timestamps from initialization to availability.	Troubleshoot build issues or understand the environment’s setup history.
ENVIRONMENT PACKAGES	Displays any additional user-installed or project-specific packages in this environment. If no custom packages are present, the tab indicates that package information is unavailable for the current workflow.	Help diagnose compatibility issues, replicate workflows, or ensure consistent project dependencies.

Related Resources

Start an AI/ML Training Run (Tutorial)

Navigate the AI/ML Hub

Submit a Help Ticket