List of supported models
- Meta Series:: Llama-3-8B-Instruct, Llama-3.2-1B-Instruct
- Microsoft Series: Phi-3-mini-4k-instruct (specially optimized version)
Extensibility note
As a result of the adoption ofadapter mechanism, which theoretically supports any Transformer-based model. Current limitations mainly come from: 1) the need to rewrite thesrc/kblam/models/Adapter code under 2) Model hidden layer dimensions need to match knowledge vectors. New model adaptations contributed by the community are being gradually merged.
Technology Roadmap
According to the GitHub discussion forum, the next step will be to support: 1) Mistral architecture 2) Chinese models such as Qwen 3) multimodal extensions. Developers can modify theKeyValueEmbedderclass to implement custom integrations.
This answer comes from the articleKBLaM: An Open Source Enhanced Tool for Embedding External Knowledge in Large ModelsThe
May not be reproduced without permission:AI productivity tools " What big language models does KBLaM currently support? Will it be extended to more models in the future?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil