qmem 0.1.3

Memory library for seamless data ingestion, storage, and retrieval with customizable embedding models.

QMem

QMem is a toolkit for vector search.
It provides a command-line interface (CLI) and a Python library for interacting with a Qdrant database.
It is designed for directness and utility, offering a guided CLI for interactive tasks and a minimal Python API for programmatic control.

---

🚀 Installation

pip install qmem

---

🛠️ Commands

🔹 init

Initializes the configuration.

CLI

qmem init

---

🔹 create

Creates a vector collection.

CLI

qmem create

Library

import qmem

qmem.create(
    collection_name="my-collection",
    dim=1536,
    distance_metric="cosine"
)

---

🔹 ingest

Ingests data into a collection.

CLI

qmem ingest

Library

import qmem

qmem.ingest(
    file="path/to/data.jsonl",
    embed_field="text"
)

---

🔹 retrieve

Performs a vector search.

CLI

qmem retrieve "your query text"

Library

import qmem

results = qmem.retrieve(
    query="your query text",
    top_k=3
)
print(results)

---

🔹 index

Creates an index on metadata for filtering.

CLI

qmem index

---

🔹 filter

Retrieves records by metadata.

CLI

qmem filter

Library

import qmem

filter_payload = {
  "must": [
    { "key": "genre", "match": { "value": "Sci-Fi" } }
  ]
}

results = qmem.filter(filter_json=filter_payload, limit=10)
print(results)

---

🔹 retrieve-filter

Combines vector search with metadata filtering.

CLI

qmem retrieve-filter "your query text"

Library

import qmem

filter_payload = {
  "must": [
    { "key": "genre", "match": { "value": "Sci-Fi" } }
  ]
}

results = qmem.retrieve_filter(
    query="your query text",
    filter_json=filter_payload,
    top_k=2
)
print(results)

---

📜 License

MIT