piramid

Python client for Piramid, an inference engine for retrieval systems.

pip install piramid

from piramid import Piramid

with Piramid("http://127.0.0.1:6333") as db:   # key from api_key= or PIRAMID_API_KEY
    db.embed("docs", ["Compaction rewrites the record store without deleted documents."])

    answer = db.generate(
        messages=[{"role": "user", "content": "What happens to deleted documents?"}],
        collection="docs",
        k=2,
    )
    print(answer["text"], answer["retrieval"]["passages"])

    for event in db.stream(prompt="Piramid is", max_new_tokens=32):
        if event["event"] == "token":
            print(event["data"]["text"], end="")

Collections: collections, create_collection, collection, delete_collection, count, compact. Documents: embed, insert, upsert, get, list, delete. Search: search (by vectors), search_text (embedded by the server), both with a metadata filter such as {"topic": {"eq": "storage"}}. Server: health, version, model.

A refused request raises PiramidError with the server's message and HTTP status. /v1/chat/completions is OpenAI-compatible, so the openai package works against it directly.

Server docs: https://piramiddb.com/docs. MIT licensed.