Skip to main content

Settings — application settings

The Settings screen centralizes DM Loremind's global settings: artificial intelligence, embeddings (RAG), language and data backup.

AI​

Configuration of the active AI provider. Details in AI Assistant → Providers.

  • Provider: five providers are available — Ollama (local), 1min.ai, OpenRouter, Mistral and Gemini (cloud).
  • Default model: one model selected per provider, used for generations.

Advanced settings​

Beyond the choice of model, a few fine-grained settings are available:

  • Context window (num_ctx) — for Ollama, defines how much text the model can take into account at once.
  • PDF import — size of the chunks split at import time, and the timeout for AI calls.

Embeddings (RAG Workshop)​

Embeddings power the semantic search of the RAG Workshop:

  • Embeddings provider: Ollama or Mistral.
  • Number of retrieved chunks (top-K): how many relevant passages are retrieved to feed the context.
  • Auto-pull of the embeddings model: automatic download of the embeddings model if needed.

Language​

The interface is available in French and English, with complete translations. The selector is in Settings → Language; the change applies on the fly, without a restart, and your choice is remembered.

Data backup​

From Settings, you can export and re-import all of your data. See Backups and restore.

Persistence​

AI and embeddings settings do not live in the database: they are kept in a configuration file on the Brain service side (data/settings.json). Any change is taken into account starting from the next request, without restarting the application.