Madhesh4124/metaphor-detection-backend

Fix function signature and docstrings in single-pass interpretation

Single-pass unified Gemini verification and suppress 5-layer output on misclassifications

Add root GET / health check endpoint for Hugging Face Spaces

Add Hugging Face Hub model download fallback and sanitize MongoDB connection URL

Fix database import path resolution for root app.py

Fix PyTorch CPU installation with extra-index-url in Dockerfile

Reworked on the UI and Implemented simple XAI interface

Update Backend with Latest Gemma model

Fix majority-vote label, add context-aware detection, caches, full Gemini prompt

Isolate history per browser session using session_id

Ping every 6hrs, add live demo URL to README

Restore Gemini interpretation, history endpoints, and sentence-level analysis

Add /predict alias endpoint for frontend compatibility

Fix Docker build: CPU-only torch, remove unnecessary deps, clean imports

Fix requirements list and cleanup Dockerfile

Add MongoDB timeout and connection heartbeat logs

Fixed bugs in Dockerfile

updated requirements.txt

Updated README.md

Final backend with lazy loading and fixed return values

Finalizing deployment: Consolidated app.py, Docker configuration, and Vercel-ready frontend

Add Hugging Face metadata to README

Update app.py to load models from HF Hub

Update project title and terminology

Deleted unwanted files

Deleted unwanted files

Added new dataset and training README

Removed outdated dataset files

Implemented Cross-Lingual Interpretation

Implemented Cross-Lingual Interpretation

Delete LICENSE

Updated backend and frontend for faster API calls

Adding Telugu dataset

Remove windsurf workflow files

Remove CI workflow

Remove MongoDB credentials from docs and env example

Initial commit: Multilingual Metaphor Detection and Interpretation System