From 809f5def0c3a49c10d41a8e2165192fe5fa938da Mon Sep 17 00:00:00 2001 From: Mitja Felicijan Date: Fri, 20 Feb 2026 14:13:15 +0100 Subject: Enhance README with project details and usage Expanded project description to include embedding models, vector database creation, and local inference details. --- README.md | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'README.md') diff --git a/README.md b/README.md index 8045942..a23854a 100644 --- a/README.md +++ b/README.md @@ -1,5 +1,14 @@ An experiment using tiny LLMs as NPCs that could be embedded into the game. +Embed models into the game, build a simple vector database from text, embed +prompts, retrieve top‑k by cosine similarity, and feed context into tiny +CPU LLMs for NPC interactions. + +**No external API calls.** Everything is local, directly using GGUF models +and [llama.cpp](https://github.com/ggml-org/llama.cpp) for inference. + +https://github.com/user-attachments/assets/863b75eb-0da7-4235-8112-f00bc82d81f6 + > [!NOTE] > This project is just for fun, to see how LLMs would fare as NPCs. Because of > the non-deterministic nature of LLMs, the results vary and are often quite -- cgit v1.2.3