Skip to content

Latest commit

 

History

History
59 lines (48 loc) · 1.59 KB

README.md

File metadata and controls

59 lines (48 loc) · 1.59 KB

dhlab-visual-resolution-syntactic-ambiguity

Semester Project Autumn 2024-2025 - A Visual Resolution for Syntactic Ambiguity

Project Structure

main/
├── interface-prototype/     # Frontend code for visual resolution
├── parsing-model/          # Parsing model code
├── LICENSE                  
└── README.md               # Project documentation

How to Run

Frontend Application

Running Full Stack (Frontend + Backend)

# Run everything with Docker from project root
docker compose up

Running Frontend Only

# Prerequisites: Install yarn if not present
brew install yarn

# Verify yarn installation
yarn --version

# Start frontend development server
cd frontend/app
yarn install
yarn run dev

Frontend will be accessible at: http://localhost:3000

Parsing Model

The model checkpoints can be downloaded from Google Drive:

  • SAPar.PTB.BERT.95.85.pt
  • SAPar.PTB.BERT.POS.95.86.pt
# Mount local directory and run container
docker run -it -v /path/to/your/local/SAPar:/app sapar bash

# Execute prediction script
python prediction.py

Prediction results will be saved to output/ directory as JSON files.

Data Sources

  • Ambiguous Data: dataset/AmbiEnt_ambiguity_category/category_syntactic.json
  • Non-ambiguous Data: dataset/MNLI_sampled_data.csv

Results Directory Structure

├── prompt_based_result/    # Results from prompt-based experiments
└── parse_result/           # Parsing results and disambiguation analysis