Other Options
Documentation Index
Fetch the complete documentation index at: https://help.hyperscience.ai/llms.txt
Use this file to discover all available pages before exploring further.
Trainer Resiliency
This feature is available in v40 and later. Trainer Resiliency allows you to automatically create checkpoints of training data and model training. The system uses these checkpoints to resume model-training tasks in the event that training is interrupted.
Automatic Transmission of Usage Data
In v39 and later, on-premise, internet-connected instances send Usage Reports (v41 and earlier) or Usage Bundles (v42 and later) to Hyperscience. These exports are sent daily and contain data on the use and configuration of Hyperscience over the past day.
Processing EML files’ attachments
In v36.0.21+, v37.0.14+, v38.1.8+, and v39 and later, by default, the system processes both EML files and their attachments. The system creates a PDF from the content of the email's body, and processes each of the file's attachments as a separate document.
In v33.1.0 and later, Hyperscience gives you the possibility to deploy the application on RHEL 8.4 or later. RHEL 8.4 and later uses Podman instead of Docker. The sections below describe how you can configure Podman.
As part of High Availability and Disaster Recovery (HA/DR) best practices, we recommend deploying the Hyperscience application on multiple machines and using a load balancer to distribute web requests. Note that the trainer does not distribute requests across multiple machines automatically.
In v30, we introduced the SDM_BLOCKS_TASK_POLL_INTERVAL ".env" file variable, which gives you more control over your instance's task-polling intervals and, in turn, your overall submission throughput. In a flow, each block includes multiple tasks that can be configured.
Hyperscience provides flexible options for getting documents into Hyperscience for processing, receiving notifications during processing, and retrieving output after processing. This is done through the use of input and output connections.
To set a time zone for your installation, set the following variable to your region, as specified in this list of TZ database time zones: SYSTEM_TIME_ZONE=
It is normal—and expected—for the Hyperscience application to consume 100% of its CPUs' resources. In v28.0.8+, v28.2.1+, and v30 and later, you can limit the percentage of CPU resources it consumes by setting the HS_MAX_CPU_PERC variable in your environment file.
We support a minimum of 8 cores for each CPU in a VM running the Hyperscience application or trainer. However, depending on the configuration of your instance and your use of Hyperscience, you may benefit from having more cores per CPU.
Proxy server setup: If proxy servers are required to make outbound HTTP or HTTPS connections in your environment, the application can be configured by setting the proxy configuration in the “.env” file. The following variables are supported: HTTP_PROXY, HTTPS_PROXY.
Overview: This article describes the security-related settings available for your Hyperscience instance: TLS / HTTPS for inbound and outbound connections, certificate management, HSTS, TLS version and cipher requirements, listening ports, server-side configuration.
You can display a custom login warning message to users when they first log in to the Hyperscience application. The message is shown in a dialog box that the user can dismiss by clicking the dialog's Continue button.
In v37.0.12 and earlier and in v38.0.1, the only supported HTML encoding type is UTF-8. We added support for Windows-1252 encoding in v37.0.13 and v38.0.2, along with the HTML_SUPPORTED_ENCODING_TYPES ".env" file variable.
Tasks Per Database Transaction
In some situations, the system may include many tasks in a single database transaction. However, if the number of tasks being created at that time is very high, delays in processing may occur.
Limiting Task Parallelism in Flow Runs
Long-running transactions, such as those created by flow runs with many branches, can result in responsiveness issues across the system. For example, a flow run containing a large number of Fork or Foreach Blocks can create more tasks or subflow runs, impacting overall performance.
If your organization develops and customizes flows, you can give your System Admins additional permissions by adding the “.env” file variables described in the sections below.
When customizing flows, you can connect one subflow to another in a “nested” manner. However, the more subflows are nested, the greater the potential for recursion-related errors.
This article describes the customization options available for processing PDF files in Hyperscience. PDF Page Box dimensions are significant for commercial printing purposes.
Reducing RAM Used by Block Processes
In v42 and later, the system does not start block processes automatically after a flow deployment unless tasks have been scheduled for the blocks.
These values of the variables described in this article should not be edited unless you are advised to do so by your Hyperscience representative.
The system runs a periodic health check on various components, including the database, file store, RAM, and trainers. The results of these health checks are shown on the System & Health page in the application.