Long-Term Memory Engine
Configure Zep Contextual Compression and Refiltering pipelines.
Contextual Compression Engine
Select the fast, lightweight model that will analyze raw Zep memory and extract only relevant facts before passing it to the main AI prompt.
Engine Status:
Recommended: Small, fast instruction-tuned models (e.g. Qwen 2.5 14B or Llama 3 8B).