Interface: LocalOnnxRerankerConfig

Defined in: Services/Search/SearchPipelineConfig.ts:109

Properties

dtype?

optional dtype?: string

Defined in: Services/Search/SearchPipelineConfig.ts:120

ONNX model quantization dtype. Default: 'fp32'.

Left at fp32 unlike the embedder: cross-encoder scores feed directly into result ordering, and the MiniLM rerankers are small enough (~90-120 MB at fp32) that quantizing saves little. Benchmark before changing.

---

modelName?

optional modelName?: string

Defined in: Services/Search/SearchPipelineConfig.ts:112

HuggingFace model name. Default: 'cross-encoder/ms-marco-MiniLM-L-6-v2'

---

provider

provider: "local-onnx"

Defined in: Services/Search/SearchPipelineConfig.ts:110