Ollama Settings

The Ollama tab manages your Local, on-device offline AI models that power dIKta.me's "Local" environment route. It ensures absolute privacy because no data is sent over the internet for processing.

Ollama is a powerful, lightweight framework running in the background of your PC. dIKta.me communicates heavily with it to format and parse your transcriptions completely privately.

Server Configuration

By default, dIKta.me assumes Ollama is running on the standard Localhost environment on your PC.

• Host URL (http://localhost:11434): If you have a dedicated server tower running Ollama elsewhere on your network, change this URL to point the LLM processor at the right IP address. Otherwise, leave it as default.
• Test Connection: Instantly verifies that your PC can actively speak with the Ollama server and lists the active installed background framework version.

Model Management Hub

Instead of forcing you to use terminal commands, the dIKta.me Ollama tab acts as a comprehensive manager that integrates directly into the pipeline:

• Installed Models: A live table of every model currently downloaded in your Ollama library, displaying file sizes and parameter structures (e.g. llama3:8b, mistral).
• Delete: Instantly free up hard drive space by deleting an installed model natively inside the GUI.
• Download New Model: Need phi3 or gemma? Type the model tag into the pull bar and dIKta.me will show you a progress overlay as the model streams and installs directly into your framework.

Once the models are installed here, you can seamlessly navigate back to the AI Engine tab and set any of these newly downloaded offline models as your primary Dictation formatter.