KNN Router

The KNN (K-Nearest Neighbors) router uses semantic similarity to match incoming interroge against a database of historical interroge with known optimal model assignments. Cela permet learned routing that improves over time.

Fonctionnement

1. Embed query -- convert the incoming query vers un vector embedding
2. KNN search -- find the K most similar past interroge in the embedding store
3. Vote -- aggregate the model assignments of the K neighbors
4. Select -- choose the model avec le most votes (weighted by similarity)

Training Data

The KNN router builds its dataset from:

• Agent session logs with quality ratings
• A/B test results from prompt evolution
• Manual feedback and corrections

Configuration

[router]
strategy = "knn"

[router.knn]
k = 5
embedding_provider = "ollama"
embedding_model = "nomic-embed-text"
min_similarity = 0.6
min_dataset_size = 100
fallback_strategy = "heuristic"

Cold Start

When insufficient training data est disponible (below min_dataset_size), the KNN router falls back vers le heuristic strategy.

Voir aussi Pages

• Router Overview
• Heuristic Router
• Automix Router
• Embeddings Memory