21min

Experiment Tracking

The library provides segmind.tracking API written in python, running on top of the MLFLow tracking. Use the library to track machine learning experiments. For Tensorflow, Keras, PyTorch, and Scikit Learn, we have special lightweight Integrations to make it fast and easy to set up experiment tracking.

All Native imports from segmind.tracking

segmind.tracking.set_project

This must be initialised at the beginning of your code/notebook, in order to use all other segmind.tracking methods. It lets you associate your further work on the code with a Segmind Project ID.

Name

Type

Default

Description

project_id*

string

none

Set the project ID. To learn more about creating a project and getting the project_id, click here.

Python
|

segmind.tracking.start_run

This must be initialised after running set_project("project_id"). This creates a unique run_uuid in our system and associates all the further tasks/artifacts/params etc with this run_uuid.

Name

Type

Default

Description

project_id*

string

none

Set the project ID. To learn more about creating a project and getting the project_id, click here.

Python
|

segmind.tracking.end_run

This will end your run, which means, your further run data will not be associated with any run_uuid, and will be lost. Only use this if you don't want to associate your params/metrices/artifacts with any run and you don't want it to be logged.

Name

Type

Default

Description

project_id*

string

none

Set the project ID. To learn more about creating a project and getting the project_id, click here.

Python
|

segmind.tracking.get_run

This will get the Current Run Info. You can call this method at anytime in your notebook - Required that your Run is initialised and not ended.
Python
|

segmind.tracking.set_runid

If you have multiple-runs in your notebook and you need your further params/metrices/artifacts to be associated with a particular run. Then you must use this. This will associate your further RUN data to this run_uuid.

Name

Type

Default

Description

run_uuid*

string

none

The unique run_uuid that was generated by start_run OR get_run() method. You can also get this from Segmind Portal.

Python
|

segmind.tracking.log_param

This will log 1-single param to your associated RUN. Although, params are logged automatically by adding segming-callbacks to your Machine Learning Models while running your .fit or training commands.

But, if you need to manually log params, then this method is very helpful. PS: If you haven't started your run by running start_run() this will automatically create a run for you. PS: key should be unique. And you won't be able to change/update the parameter value, if it's set once. PS: You won't be able to see your params on Segmind Portal, until and unless you have a metric associated with your run - which is done by log_metric() method. PS: You can log UNLIMITED Params, as required in your RUN.

Name

Type

Default

Description

key*

string

none

PARAMETER

value*

string

none

RUN VALUE

Python
|

segmind.tracking.log_params

This will log multiple params to your associated RUN.

This is similar to log_param method.

Name

Type

Default

Description

key*

string

none

PARAMETER

value*

string

none

RUN VALUE

Python
|

segmind.tracking.log_metric

Name

Type

Default

Description

key*

string

none



value*

string

none



Python
|

segmind.tracking.log_metrics

Python
|

segmind.tracking.log_artifact

Name

Type

Default

Description

key

string

none



value

string

none



Python
|

segmind.tracking.log_image

This will log an image to your associated Run. You can see your image in your Segmind Portal, in your associated run.

Name

Type

Default

Description

key*

string

none

unique-key for your image

image*

string

none

image name, what is in your notebook

Python
|

segmind.tracking.log_table

Name

Type

Default

Description

key*

string

string



table*

table

dataframe



Python
|





Updated 24 May 2022
Did this page help?
Yes
No