README.md

metadata

title: Classifier General
emoji: 🌍
colorFrom: gray
colorTo: purple
sdk: docker
pinned: false
license: mit
short_description: classifier-general

Classifier General API (Refactored)

Refactored into a modular FastAPI backend with clear layers:

• app/routers
• app/services
• app/pipelines
• app/schemas
• app/models
• app/core

Preserved Endpoint Contract

• POST /api/classifier -> returns label string
• POST /api/language -> returns language string
• POST /api/transformer -> returns { filename, content }
• POST /classify -> returns { label, language, type? }
• POST /configlabel -> returns labels array
• GET /labels -> returns labels array

POST /configlabel exact payload:

• body accepts {"labels":["label1","label2","label3"]}
• all resulting labels are trimmed, empty values removed
• duplicates are kept if they are provided
• returns the stored string[] labels

Additional operational endpoints:

• GET /health/liveness
• GET /health/readiness
• GET /endpoint/

Environment

Copy and edit:

cp .env.example .env

Key vars:

• CLASSIFIER_MODEL
• ENABLE_MODEL_QUANTIZATION
• HUGGINGFACE_TOKEN
• CLASSIFIER_ENTAILMENT_LABEL_ID (optional override when model config has no entailment label name)
• DEFAULT_LABELS_CSV

Local Run

pip install -r requirements.txt
uvicorn main:app --host 0.0.0.0 --port 4002 --reload

Docker Run

docker compose up --build

Tests

pytest -q

Notes

• OCR requires tesseract-ocr (installed in Dockerfile).
• Supported extraction formats in this refactor: .pdf, .docx, .xlsx, image formats, and plain text files.
• The classifier model is loaded directly from Hugging Face Hub and runs true zero-shot classification over runtime labels.
• Language detection runs locally via langdetect (no remote language endpoint dependency).
• /classify uses only the first PDF page for classification; /api/transformer still extracts full content.