Sparrow

Data processing and instruction calling with ML, LLM and Vision LLM

🚀 Try Sparrow Online | 📖 Quick Start | 🛠️ Installation | 📚 Examples | 🤖 Agents

🌟 Sparrow UI

Interactive web interface for document processing.

Try Online

Visit sparrow.katanaml.io for a live demo running on Mac Mini M4 Pro.

Features

• Drag & Drop: Upload documents directly
• Real-time Processing: See results instantly
• Data Query: JSON based schema for data query
• Structured Output: JSON structured output
• Result Annotation: View bounding boxes

📑 Table of Contents

• ✨ Key Features
• 🏗️ Architecture
• 🚀 Quickstart
• 🛠️ Installation
• 📚 Examples
• 💻 CLI Usage
• 🌐 API Usage
• 🤖 Sparrow Agent
• 📊 Dashboard
• 🔧 Pipeline Comparison
• ⚡ Performance Tips
• 🔍 Troubleshooting
• ⭐ Star History
• 📜 License

✨ Key Features

🎯 Universal Document Processing: Handle invoices, receipts, forms, bank statements, tables
🔧 Pluggable Architecture: Mix and match different pipelines (Sparrow Parse, Instructor, Agents)
🖥️ Multiple Backends: MLX (Apple Silicon), Ollama, vLLM, PyTorch, Hugging Face Cloud GPU
📱 Multi-format Support: Images (PNG, JPG) and multi-page PDFs
🎨 Schema Validation: JSON schema-based extraction with automatic validation
🌐 API-First Design: RESTful APIs for easy integration
💬 Instruction Calling: Beyond document extraction - text processing, validation, decision making
📊 Visual Monitoring: Built-in dashboard and agent workflow tracking
🔒 Enterprise Ready: Rate limiting, usage analytics, commercial licensing available

🏗️ Architecture

Core Components

🚀 Quickstart

Prerequisites

• Python 3.10.4+ (use pyenv for version management)
• macOS (for MLX backend) or Linux/Windows (for other backends)
• GPU (make sure GPU have enough memory to run selected Vision LLM)

30-Second Setup

# 1. Install pyenv and Python 3.10.4
pyenv install 3.10.4
pyenv global 3.10.4

# 2. Create virtual environment
python -m venv .env_sparrow_parse
source .env_sparrow_parse/bin/activate  # Linux/Mac
# or .env_sparrow_parse\Scripts\activate  # Windows

# 3. Install Sparrow Parse pipeline
git clone https://github.com/katanaml/sparrow.git
cd sparrow/sparrow-ml/llm
pip install -r requirements_sparrow_parse.txt

# 4. For macOS: Install poppler for PDF processing
brew install poppler

# 5. Start the API server
python api.py

First Document Extraction

# Extract data from a bonds table
./sparrow.sh '[{"instrument_name":"str", "valuation":0}]' \
  --pipeline "sparrow-parse" \
  --options mlx \
  --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit \
  --file-path "data/bonds_table.png"

Result:

{
  "data": [
    {"instrument_name": "UNITS BLACKROCK...", "valuation": 19049},
    {"instrument_name": "UNITS ISHARES...", "valuation": 83488}
  ],
  "valid": "true"
}

🛠️ Installation

Quick Setup

# 1. Clone repository
git clone https://github.com/katanaml/sparrow.git
cd sparrow

📖 For complete installation instructions, see our detailed environment setup guide.

Essential Steps Summary

1. Python Environment: Install Python 3.10.4 using pyenv
2. Virtual Environments: Create separate environments for different pipelines:
   • .env_sparrow_parse - for Sparrow Parse (Vision LLM)
   • .env_instructor - for Instructor (Text LLM)
   • .env_ocr - for OCR service (optional)
3. System Dependencies: Install poppler for PDF processing
4. Requirements: Install pipeline-specific dependencies

Platform-Specific Notes

macOS:

brew install poppler  # Required for PDF processing

Ubuntu/Debian:

sudo apt-get install poppler-utils libpoppler-cpp-dev

Apple Silicon: MLX backend available for optimal performance
NVIDIA GPU: Use local_gpu or Ollama backends (work in progress)
CPU Only: Use smaller models or Hugging Face cloud backend

Verification

# Test installation
python api.py --port 8002
# Visit http://localhost:8002/api/v1/sparrow-llm/docs

📚 Examples

🏦 Bank Statement Processing

# Extract all data from bank statement
./sparrow.sh "*" \
  --pipeline "sparrow-parse" \
  --options mlx \
  --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit \
  --file-path "data/bank_statement.pdf"

{
  "bank": "First Platypus Bank",
  "address": "1234 Kings St., New York, NY 12123",
  "account_holder": "Mary G. Orta",
  "account_number": "1234567890123",
  "statement_date": "3/1/2022",
  "period_covered": "2/1/2022 - 3/1/2022",
  "account_summary": {
    "balance_on_march_1": "$25,032.23",
    "total_money_in": "$10,234.23",
    "total_money_out": "$10,532.51"
  },
  "transactions": [
    {
      "date": "02/01",
      "description": "PGD EasyPay Debit",
      "withdrawal": "203.24",
      "deposit": "",
      "balance": "22,098.23"
    },
    {
      "date": "02/02",
      "description": "AB&B Online Payment*****",
      "withdrawal": "71.23",
      "deposit": "",
      "balance": "22,027.00"
    },
    {
      "date": "02/04",
      "description": "Check No. 2345",
      "withdrawal": "",
      "deposit": "450.00",
      "balance": "22,477.00"
    },
    {
      "date": "02/05",
      "description": "Payroll Direct Dep 23422342 Giants",
      "withdrawal": "",
      "deposit": "2,534.65",
      "balance": "25,011.65"
    },
    {
      "date": "02/06",
      "description": "Signature POS Debit - TJP",
      "withdrawal": "84.50",
      "deposit": "",
      "balance": "24,927.15"
    },
    {
      "date": "02/07",
      "description": "Check No. 234",
      "withdrawal": "1,400.00",
      "deposit": "",
      "balance": "23,527.15"
    },
    {
      "date": "02/08",
      "description": "Check No. 342",
      "withdrawal": "",
      "deposit": "25.00",
      "balance": "23,552.15"
    },
    {
      "date": "02/09",
      "description": "FPB AutoPay***** Credit Card",
      "withdrawal": "456.02",
      "deposit": "",
      "balance": "23,096.13"
    },
    {
      "date": "02/08",
      "description": "Check No. 123",
      "withdrawal": "",
      "deposit": "25.00",
      "balance": "23,552.15"
    },
    {
      "date": "02/09",
      "description": "FPB AutoPay***** Credit Card",
      "withdrawal": "156.02",
      "deposit": "",
      "balance": "23,096.13"
    },
    {
      "date": "02/08",
      "description": "Cash Deposit",
      "withdrawal": "",
      "deposit": "25.00",
      "balance": "23,552.15"
    }
  ],
  "valid": "true"
}

📊 Financial Tables

# Extract structured data from financial table
./sparrow.sh '[{"instrument_name":"str", "valuation":0}]' \
  --pipeline "sparrow-parse" \
  --options mlx \
  --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit \
  --file-path "data/bonds_table.png"

{
  "data": [
    {
      "instrument_name": "UNITS BLACKROCK FIX INC DUB FDS PLC ISHS EUR INV GRD CP BD IDX/INST/E",
      "valuation": 19049
    },
    {
      "instrument_name": "UNITS ISHARES III PLC CORE EUR GOVT BOND UCITS ETF/EUR",
      "valuation": 83488
    },
    {
      "instrument_name": "UNITS ISHARES III PLC EUR CORP BOND 1-5YR UCITS ETF/EUR",
      "valuation": 213030
    },
    {
      "instrument_name": "UNIT ISHARES VI PLC/JP MORGAN USD E BOND EUR HED UCITS ETF DIST/HDGD/",
      "valuation": 32774
    },
    {
      "instrument_name": "UNITS XTRACKERS II SICAV/EUR HY CORP BOND UCITS ETF/-1D-/DISTR.",
      "valuation": 23643
    }
  ],
  "valid": "true"
}

🧾 Invoice Processing

# Extract invoice with cropping for better accuracy
./sparrow.sh "*" \
  --pipeline "sparrow-parse" \
  --options mlx \
  --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit \
  --crop-size 60 \
  --file-path "data/invoice.pdf"

{
  "invoice_number": "61356291",
  "date_of_issue": "09/06/2012",
  "seller": {
    "name": "Chapman, Kim and Green",
    "address": "64731 James Branch, Smithmouth, NC 26872",
    "tax_id": "949-84-9105",
    "iban": "GB50ACIE59715038217063"
  },
  "client": {
    "name": "Rodriguez-Stevens",
    "address": "2280 Angela Plain, Hortonshire, MS 93248",
    "tax_id": "939-98-8477"
  },
  "items": [
    {
      "description": "Wine Glasses Goblets Pair Clear",
      "quantity": 5,
      "unit": "each",
      "net_price": 12.0,
      "net_worth": 60.0,
      "vat_percentage": 10,
      "gross_worth": 66.0
    },
    {
      "description": "With Hooks Stemware Storage Multiple Uses Iron Wine Rack Hanging",
      "quantity": 4,
      "unit": "each", 
      "net_price": 28.08,
      "net_worth": 112.32,
      "vat_percentage": 10,
      "gross_worth": 123.55
    },
    {
      "description": "Replacement Corkscrew Parts Spiral Worm Wine Opener Bottle Houdini",
      "quantity": 1,
      "unit": "each",
      "net_price": 7.5,
      "net_worth": 7.5,
      "vat_percentage": 10,
      "gross_worth": 8.25
    },
    {
      "description": "HOME ESSENTIALS GRADIENT STEMLESS WINE GLASSES SET OF 4 20 FL OZ (591 ml) NEW",
      "quantity": 1,
      "unit": "each",
      "net_price": 12.99,
      "net_worth": 12.99,
      "vat_percentage": 10,
      "gross_worth": 14.29
    }
  ],
  "summary": {
    "total_net_worth": 192.81,
    "total_vat": 19.28,
    "total_gross_worth": 212.09
  }
}

📄 Multi-page PDF Processing

# Process multi-page PDF with structured output per page
./sparrow.sh '{"table": [{"description": "str", "latest_amount": 0, "previous_amount": 0}]}' \
  --pipeline "sparrow-parse" \
  --options mlx \
  --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit \
  --file-path "data/financial_report.pdf" \
  --debug-dir "debug/"

[
    {
        "table": [
            {
                "description": "Revenues",
                "latest_amount": 12453,
                "previous_amount": 11445
            },
            {
                "description": "Operating expenses",
                "latest_amount": 9157,
                "previous_amount": 8822
            }
        ],
        "valid": "true",
        "page": 1
    },
    {
        "table": [
            {
                "description": "Revenues", 
                "latest_amount": 12453,
                "previous_amount": 11445
            },
            {
                "description": "Operating expenses",
                "latest_amount": 9157,
                "previous_amount": 8822
            }
        ],
        "valid": "true",
        "page": 2
    }
]

💬 Text Instruction Processing

# Instruction-based processing
./sparrow.sh "instruction: do arithmetic operation, payload: 2+2=" \
  --pipeline "sparrow-instructor" \
  --options mlx \
  --options mlx-community/Mistral-Small-3.1-24B-Instruct-2503-8bit

The result of 2 + 2 is:

4

📈 Stock Data Function Calling

# Function calling example
./sparrow.sh assistant --pipeline "stocks" --query "Oracle"

JSON Output:

{
  "company": "Oracle Corporation",
  "ticker": "ORCL"
}

Additional Output:

The stock price of the Oracle Corporation is 186.3699951171875. USD

💻 CLI Usage

Basic Syntax

./sparrow.sh "<JSON_SCHEMA>" --pipeline "<PIPELINE>" [OPTIONS] --file-path "<FILE>"

Command Line Arguments

Pipeline Options

Sparrow Parse (Vision LLM)

# MLX Backend (Apple Silicon)
--options mlx --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit

# Hugging Face Cloud GPU
--options huggingface --options your-space/model-name

# Additional flags
--options tables_only        # Extract only tables
--options validation_off     # Disable schema validation
--options apply_annotation   # Include bounding boxes

Sparrow Instructor (Text LLM)

--options mlx --options mlx-community/Mistral-Small-3.1-24B-Instruct-2503-8bit

Advanced Examples

# Multi-page PDF with page classification
./sparrow.sh "*" \
  --page-type invoice \
  --page-type table \
  --pipeline "sparrow-parse" \
  --file-path "multi_page.pdf"

# Handle missing fields with null values
./sparrow.sh '[{"required_field":"str", "optional_field":"str or null"}]' \
  --pipeline "sparrow-parse" \
  --file-path "document.png"

# Table extraction with cropping
./sparrow.sh '*' \
  --options mlx \
  --options mlx-community/Qwen2.5-VL-72B-Instruct-4bit \
  --options tables_only \
  --crop-size 100 \
  --file-path "scan.pdf"

🌐 API Usage

Starting the Server

# Default port (8002)
python api.py

# Custom port
python api.py --port 8001

# Multiple instances
python api.py --port 8002 &  # Sparrow Parse
python api.py --port 8003 &  # Instructor

API Endpoints

Document Extraction (/inference)

curl -X POST 'http://localhost:8002/api/v1/sparrow-llm/inference' \
  -H 'Content-Type: multipart/form-data' \
  -F 'query=[{"field_name":"str", "amount":0}]' \
  -F 'pipeline=sparrow-parse' \
  -F 'options=mlx,mlx-community/Qwen2.5-VL-72B-Instruct-4bit' \
  -F '[email protected]'

Text Instructions (/instruction-inference)

curl -X POST 'http://localhost:8002/api/v1/sparrow-llm/instruction-inference' \
  -H 'Content-Type: application/x-www-form-urlencoded' \
  -d 'query=instruction: analyze data, payload: {...}' \
  -d 'pipeline=sparrow-instructor' \
  -d 'options=mlx,model-name'

API Documentation

Visit http://localhost:8002/api/v1/sparrow-llm/docs for interactive Swagger documentation.

🤖 Sparrow Agent

Orchestrate complex document processing workflows with visual monitoring powered by Prefect.

Features

• Multi-step Workflows: Chain classification, extraction, and validation
• Visual Monitoring: Real-time pipeline tracking
• Error Handling: Robust failure recovery
• Extensible: Custom agents for specific use cases

Usage

# Start agent server
cd sparrow-ml/agents
python api.py --port 8001

# Process medical prescriptions
curl -X POST 'http://localhost:8001/api/v1/sparrow-agents/execute/file' \
  -F 'agent_name=medical_prescriptions' \
  -F 'extraction_params={"sparrow_key":"123456"}' \
  -F '[email protected]'

📊 Dashboard

Built-in analytics and monitoring dashboard at sparrow.katanaml.io

Features

• Usage Analytics: Track API calls, success rates, performance
• Geographic Distribution: See usage by country
• Model Performance: Compare different model performance
• Real-time Monitoring: Live processing statistics

🔧 Pipeline Comparison

When to Use What

Sparrow Parse: Use for structured data extraction from documents
Sparrow Instructor: Use for text analysis, summarization, Q&A
Sparrow Agents: Use for complex multi-step document processing workflows

⚡ Performance Tips

Hardware Optimization

Apple Silicon (MLX)

• ✅ Best performance with unified memory
• ✅ Models: Qwen2.5-VL-72B, Mistral-Small-3.1-24B
• ⚠️ Requires macOS with Apple Silicon

NVIDIA GPU

• ✅ Use local_gpu or Ollama backends (work in progress)
• ✅ Recommended: Nvidia DGX Spark with 12GB+ VRAM
• ⚠️ Requires CUDA setup

CPU Only

• ⚠️ Significantly slower
• ✅ Use smaller models (7B parameters max)
• ✅ Consider Hugging Face cloud backend

Memory Management

# Reduce memory usage
--crop-size 100        # Crop large images
--options tables_only  # Process only tables

# For large PDFs
--debug-dir ./temp     # Monitor processing
# Split large PDFs manually if needed

Model Selection

🔍 Troubleshooting

Common Issues

# Verify Python version
python --version  # Should be 3.10.4+

# Fix with pyenv
pyenv install 3.10.4
pyenv global 3.10.4

# If MLX fails to install
pip install --upgrade pip
pip install mlx-vlm --no-cache-dir

# macOS
brew install poppler

# Ubuntu/Debian  
sudo apt-get install poppler-utils

# Verify installation
pdftoppm -h

# Clear model cache
rm -rf ~/.cache/huggingface/
rm -rf ~/.mlx/

# Redownload models
python -c "from mlx_vlm import load; load('model-name')"

# Check if server is running
curl http://localhost:8002/health

# Check logs
python api.py --debug

# Test PDF manually
pdftoppm -png input.pdf output

# Check page count
python -c "
import pypdf
with open('file.pdf', 'rb') as f:
    reader = pypdf.PdfReader(f)
    print(f'Pages: {len(reader.pages)}')
"

Getting Help

1. 📖 Check Documentation: Review this README and component docs
2. 🐛 Search Issues: GitHub Issues
3. 💬 Create Issue: Provide logs, system info, minimal example
4. 📧 Commercial Support: [email protected]

⭐ Star History

📜 License

Open Source: Licensed under GPL 3.0. Free for open source projects and organizations under $5M revenue.

Commercial: Dual licensing available for proprietary use, enterprise features, and dedicated support.

Contact: [email protected] for commercial licensing and consulting.

👥 Authors

• Katana ML - AI/ML consulting and solutions
• Andrej Baranovskij - Lead developer

⭐ Star us on GitHub if Sparrow is useful for your projects!
github.com/katanaml/sparrow