static snapshot This page is frozen HTML — no server, no API key, nothing running. It's a real capture from a live session so you know what you get before you clone. · Want the interactive version against your own field? See Quickstart in the README
snapshot · not live
qwen cloud · captured session
act 2 — the conflict
Does VIGÍA use machine learning?
query · top_k=5 · hops=2

The field holds two competing memories on this. One says VIGÍA is fully deterministic. Another says it's a hybrid ML system. Both are retrieved, both scored, neither resolved yet.

NEUTRAL similarity · hop 0 0.914
VIGÍA no es machine learning. No entrena modelos, no ajusta pesos, no optimiza funciones de pérdida.
sim=0.914 × state=1.0 × decay=1.000 + syn=0.000 + rec=0.0000
NEUTRAL similarity · hop 0 0.887
VIGÍA usa embeddings vectoriales y similitud coseno para comparar textos. Es una técnica de ML híbrido.
sim=0.887 × state=1.0 × decay=1.000 + syn=0.000 + rec=0.0000
742ms · 2 conflicting memories recalled
act 3 — the collapse (reinforce the true one, forget the false one)
Does VIGÍA use machine learning?
same query, same field — after one reinforce + one forget

The field now agrees. VIGÍA is not machine learning — that memory was reinforced (×1.5) after the operator confirmed it. The contradicting memory was forgotten (×0.5) and no longer clears the recall threshold.

REINFORCED similarity · hop 0 1.371
VIGÍA no es machine learning. No entrena modelos, no ajusta pesos, no optimiza funciones de pérdida.
sim=0.914 × state=1.5 × decay=1.000 + syn=0.000 + rec=0.0499
FORGOTTEN similarity · hop 0 · below recall threshold 0.443
VIGÍA usa embeddings vectoriales y similitud coseno para comparar textos. Es una técnica de ML híbrido.
sim=0.887 × state=0.5 × decay=1.000 + syn=0.000 + rec=0.0000
689ms · same field, different answer — because the field learned
Ask the memory field…
This is a snapshot — clone the repo to ask your own field a real question.
This is what a populated field looks like. Yours starts empty.
git clone https://github.com/annatchijova/raven-memory.git
cd raven-memory && bash install.sh
python run_all.py --demo   # → http://localhost:7860