Model Management

Model Management in NELA is handled from Settings and mode selectors. It covers model download, removal, runtime tuning, and compatibility checks.

How it works

Models are grouped by task class: LLM, Vision, TTS, ASR, Embedding, Classifier, and Grader.
Downloads and uninstall operations are managed from the in-app settings workflow.
Compatibility estimation checks RAM, CPU, and disk before large downloads.
Runtime parameters (context size, max tokens, temperature, top-p/top-k, repeat penalty, and backend controls) are configurable per model.
Mode selectors let you switch active installed models without changing conversation flow.

Rendering diagram...