@knolo/semantic-ollama

Optional Ollama embedding provider for Knolo’s hybrid retrieval layer.

npm install @knolo/semantic-ollama @knolo/core

import { OllamaEmbeddingProvider } from '@knolo/semantic-ollama';

const embeddings = new OllamaEmbeddingProvider({
  endpoint: 'http://localhost:11434',
  modelId: 'qwen3-embedding:4b',
});
const vector = await embeddings.embedQuery('billing policy');

The provider calls a local Ollama server through its /api/embeddings endpoint. Semantic reranking is optional; Knolo’s default path remains local, lexical, and deterministic. V5 Knowledge Image verification and migration are provided by @knolo/core.

License

Apache-2.0