pyproject.toml

[project]
name = "german-document-classifier"
version = "0.1.0"
description = "German document classifier using BERT and fastapi"
readme = "README.md"
requires-python = ">=3.10"
dependencies = [
    #  Web API
    "fastapi>=0.124.0",
    "uvicorn>=0.38.0",
    "python-multipart>=0.0.20", # For file uploads
    #  Machine Learning / Transformers
    "torch>=2.5.0",             # Explicitly add torch
    "transformers>=4.40.0",
    "scikit-learn>=1.7.2",
    "accelerate>=1.11.0",       # optimizing inference
    "evaluate>=0.4.2",          # metrics
    #  Document Processing / OCR
    "PyMuPDF>=1.26.6",          # PDF extraction
    "pytesseract>=0.3.13",      # OCR
    "pillow-heif>=1.1.1",       # HEIC Image support
    "python-docx>=1.2.0",       # Word docs
    "puremagic>=1.30",          # Mime type detection
    #  Utilities
    "tqdm>=4.66.5",             # Progress bars
    "wandb>=0.18.5",            # Logging experiments
    "sentry-sdk>=2.43.0",       # Error tracking             # data generation for
    "faker>=38.2.0",
    "seaborn>=0.13.2",
]

[tool.setuptools]
packages = ["app"]