Skip to main content
Version: v1.3.0

App settings

Overview​

H2O Hydrogen Torch uses specific default values for certain settings you use when defining the settings to import a dataset or when defining an experiment. For example:

  • When importing a dataset, the default S3 bucket name is h2o-release/hydrogen-torch/* (for example, h2o-release/hydrogen-torch/1.3.0)
  • When determining a value for a slider setting, the slider has a defined maximum value

These settings using specific default or maximum values are referred to as application settings. Application settings are grouped into the following sections:

Define app settings​

To define a default value for one or all of the application settings, consider the following instructions:

  1. In the H2O Hydrogen Torch navigation menu, click Settings.
  2. In the Settings card, select an alternative default value for one or all of the settings.
  3. Click Save settings persistently.

Restore default app settings​

If you want to restore the default setting values of H2O Hydrogen Torch, consider the following instructions:

  1. In the H2O Hydrogen Torch navigation menu, click Settings.
  2. Click Restore default settings.

Load defined app settings​

In the case where you restored the default values but now want to use your defined application value settings (again), consider the following instructions:

  1. In the H2O Hydrogen Torch navigation menu, click Settings.
  2. Click Load settings.

Appearance settings​

Dark mode​

Enables and disables Dark mode as a theme for H2O Hydrogen Torch.

Demo mode​

Enables and disables a demo mode that provides the ability to view H2O Hydrogen Torch simplistically. Enabling Demo mode allows you to view pre-run (non-functional) datasets and experiments for all supported problem types.

Note

Demo mode requires an internet connection to be available.

Delete dialogs​

Determines whether a dialog is displayed to confirm the deletion of a dataset, experiment, project, or deployment. If this setting is turned Off, no dialog confirmation is displayed, and a dataset, experiment, project, or deployment is deleted immediately.

Maximum chart points​

Defines the maximum number of points each experiment chart (plots) renders. Charts are sub-sampled with the specified value.

Default connector settings​

Warning

Do not store your credentials on public machines. Anyone with access to that machine can use them.

AWS S3 bucket name​

Defines the default AWS bucket name H2O Hydrogen Torch references to retrieve datasets stored in an S3 bucket. The S3 bucket should include relative paths.

AWS access key​

Defines the default AWS access key H2O Hydrogen Torch uses to access the defined S3 bucket.

AWS secret key​

Defines the default AWS secret key H2O Hydrogen Torch uses to access the defined S3 bucket.

Azure Datalake connection string​

Defines the Azure Datalake string connection for dataset import that H2O Hydrogen Torch utilizes to connect to Azure Datalake.

Azure Datalake container name​

Defines the name of the Azure Datalake container H2O Hydrogen Torch connects to.

Kaggle username​

Defines the default Kaggle username H2O Hydrogen Torch references to retrieve your datasets in Kaggle.

Kaggle secret key​

Defines the default Kaggle secret key that H2O Hydrogen Torch uses to access datasets in Kaggle.

Default experiment settings​

Experience level​

When initiating an experiment, you can define your experience level with data science and machine learning, which defines the settings (hyperparameters) H2O Hydrogen Torch displays for the experiment you are initiating.

Number of workers​

Defines the default maximum number of workers the Number of workers setting makes available. The Number of workers setting is available when defining the settings for an experiment.

Mixed precision training​

Determines whether to utilize mixed precision for training, resulting in faster training times.

Mixed precision inference​

Determines whether to utilize mixed precision for inference, resulting in faster training times.

Logger​

Defines the default logger type that H2O Hydrogen Torch uses for the logger setting.

Neptune project​

Defines the default Neptune project H2O Hydrogen Torch uses for the Neptune project setting. The Neptune project setting is available when defining the settings for an experiment.

Neptune API token​

Defines the default Neptune API token H2O Hydrogen Torch uses for the Neptune API token setting. The Neptune API token setting is available when defining the settings for an experiment.

Build scoring pipelines​

After completing an experiment, this setting determines whether H2O Hydrogen Torch should build a model's H2O MLOps and Python scoring pipelines. Pipelines are built on demand if this setting is Off.

Experiment maximum settings​

Image width​

Defines the maximum default value H2O Hydrogen Torch uses for the Image width slider setting.

Image height​

Defines the maximum default value H2O Hydrogen Torch uses for the Image height slider setting.

Number of epochs​

Defines the maximum default value H2O Hydrogen Torch uses for the Epochs slider setting.

Batch size​

Defines the maximum default value H2O Hydrogen Torch uses for the Batch size slider setting.

Token length​

Defines the maximum default value H2O Hydrogen Torch uses for the Token length slider setting for a Natural Language Processing (NLP) experiment.

Gradient clip​

Determines the maximum value H2O Hydrogen Torch utilizes for the gradient clip slider available when defining an experiment's settings.

Deployment settings​

MLOps API URL​

Defines (specifies) the H2O MLOps API URL H2O Hydrogen Torch is connected.

Allow custom MLOps limits​

Determines whether H2O Hydrogen Torch displays specific settings around resource requests and limits when specifying the deployment settings of a model you want to deploy to H2O MLOps.

Feedback