Skip to content

README.md

Latest commit

 

History

History
115 lines (92 loc) · 3.58 KB

README.md

File metadata and controls

115 lines (92 loc) · 3.58 KB
Raw

Bulk Ingestion CLI — Quick Guide

  • Purpose: Ease bulk upload of large datasets to the ingestor server in sequential batches with professional logging and task polling.
  • Scope: Can optionally create and/or delete collections via flags; otherwise ensure the target collection already exists.

Prerequisites:

  • Ingestor server running (default http://localhost:8082).
  • Target collection created (see ingestion notebook/API if needed).

Install:

pip install -r scripts/requirements.txt

Run:

python scripts/batch_ingestion.py \
  --folder data/multimodal/ \
  --collection-name my_collection \
  --ingestor-host localhost \
  --ingestor-port 8082 \
  --upload-batch-size 100 \
  -v

Create/Delete collection (optional)

  • Create the collection before uploading:
python scripts/batch_ingestion.py \
  --folder data/multimodal/ \
  --collection-name my_collection \
  --create_collection
  • Delete the collection after the run completes:
python scripts/batch_ingestion.py \
  --folder data/multimodal/ \
  --collection-name my_collection \
  --delete_collection
  • Create at start and delete at end (useful for temporary runs):
python scripts/batch_ingestion.py \
  --folder data/multimodal/ \
  --collection-name my_collection \
  --create_collection --delete_collection

Notes:

  • Idempotent: Lists existing documents and skips already ingested files.
  • Progress: Uploads sequential batches and shows progress (e.g., “Uploading batch 2/43 …”).
  • Polling: Polls each ingestion task until it finishes before proceeding; fails the batch on FAILED/UNKNOWN/timeout.

Retriever API CLI — Quick Guide

Prerequisite: Before running any scripts in this folder, deploy and start the services by following the Quickstart guide. Once services are up (RAG server and Ingestor server), proceed with the commands below.

Use scripts/retriever_api_usage.py to talk to the RAG server.

  • Install dependencies: pip install -r scripts/requirements.txt
  • Default server: http://localhost:8081

Generate (default)

Generates an answer from your query and streams it to the terminal.

python scripts/retriever_api_usage.py "What is RAG?"

Search

Retrieves the top documents related to your query and returns them as JSON.

python scripts/retriever_api_usage.py --mode search "Tell me about Robert Frost's poems"

Specify collection name

Pass a collection name with your query (overrides the default multimodal_data).

# Generate
python scripts/retriever_api_usage.py \
  --payload-json '{"collection_names":["my_collection"]}' \
  "What is lion doing?"

# Search
python scripts/retriever_api_usage.py \
  --mode search \
  --payload-json '{"collection_names":["my_collection"]}' \
  "Tell me about Robert Frost's poems"

Use a payload

Customize parameters or provide messages via JSON.

# JSON string
python scripts/retriever_api_usage.py \
  --payload-json '{"messages":[{"role":"user","content":"Your query here"}]}'

# From file
python scripts/retriever_api_usage.py --mode search --payload-file scripts/payloads/search.json

Save output / change host

Save the response to a file or point to a different server URL.

python scripts/retriever_api_usage.py --mode search --output-json out/result.json "my query"
python scripts/retriever_api_usage.py --host http://my-host:8081 "my query"

Notes:

  • Provide a query or include it in the payload. For generate without a query, your payload must include messages.
  • Only fields you provide in the payload override defaults.