Overview

MemWire generates vector embeddings to store and retrieve memories. Configure the embedding model via MemWireConfig:

from memwire import MemWire, MemWireConfig

config = MemWireConfig(
    model_name="sentence-transformers/all-MiniLM-L6-v2",
    embedding_dim=384,
)
memory = MemWire(config=config)

Supported embedding backends

FastEmbed

Local on-device embeddings via Qdrant’s FastEmbed library. Supports dense and sparse models. Recommended default.

Two model types are used:

Type	Purpose
Dense model	Semantic similarity — powers memory search and recall
Sparse model	Keyword matching — used alongside dense vectors in hybrid search

Default models

from memwire import MemWire, MemWireConfig

# These are the defaults — no config needed unless you want to change them
config = MemWireConfig(
    model_name="sentence-transformers/all-MiniLM-L6-v2",
    embedding_dim=384,
    sparse_model_name="prithivida/Splade_PP_en_v1",
)
memory = MemWire(config=config)

Model	Type	Dimensions	Notes
`sentence-transformers/all-MiniLM-L6-v2`	Dense	384	Fast, multilingual-friendly, good general-purpose baseline
`prithivida/Splade_PP_en_v1`	Sparse	—	SPLADE sparse model for hybrid BM25+vector search

Changing the dense model

You can swap the dense model for any FastEmbed-compatible model. Make sure embedding_dim matches the model’s output dimension.

config = MemWireConfig(
    model_name="BAAI/bge-small-en-v1.5",
    embedding_dim=384,
)

config = MemWireConfig(
    model_name="BAAI/bge-base-en-v1.5",
    embedding_dim=768,
)

config = MemWireConfig(
    model_name="BAAI/bge-large-en-v1.5",
    embedding_dim=1024,
)

If you change the embedding model after data has already been stored, the existing vectors will be incompatible with new embeddings. Reset your Qdrant storage when switching models.

Disabling hybrid search

If you want to use dense-only search (no sparse model loaded), disable hybrid search:

config = MemWireConfig(
    use_hybrid_search=False,
)

Reranking

Enable a cross-encoder reranker to re-score the top candidates after initial retrieval. This improves precision at the cost of slightly higher latency.

config = MemWireConfig(
    use_reranking=True,
    reranker_model_name="Xenova/ms-marco-MiniLM-L-6-v2",
)

Configuration reference

Parameter	Default	Description
`model_name`	`sentence-transformers/all-MiniLM-L6-v2`	Dense embedding model. Must be FastEmbed-compatible.
`embedding_dim`	`384`	Output dimension of the dense model. Must match the model.
`sparse_model_name`	`prithivida/Splade_PP_en_v1`	Sparse model for hybrid BM25+vector search.
`use_hybrid_search`	`True`	Combine dense and sparse vectors. Disable to use dense-only search.
`use_reranking`	`False`	Apply a cross-encoder reranker to re-score top candidates.
`reranker_model_name`	`Xenova/ms-marco-MiniLM-L-6-v2`	Cross-encoder reranker model (used when `use_reranking=True`).
`embedding_cache_maxsize`	`10000`	LRU cache size for embedding vectors. Increase for large-scale workloads.

Getting started

Features

Configuration

Supported embedding backends

FastEmbed

Default models

Changing the dense model

Disabling hybrid search

Reranking

Configuration reference

​Supported embedding backends

FastEmbed

​Default models

​Changing the dense model

​Disabling hybrid search

​Reranking

​Configuration reference

Supported embedding backends

Default models

Changing the dense model

Disabling hybrid search

Reranking

Configuration reference