Skip to content

Latest commit

 

History

History
56 lines (49 loc) · 790 Bytes

README.md

File metadata and controls

56 lines (49 loc) · 790 Bytes

multi-modal-qg

A Multi-Modal Question Generation Model.

Steps to prepare data

First change pwd to dataset/

cd dataset
  1. Assign question ids
    python assign_question_id.py
    
  2. Get audio files from video files
    python get_audio.py
    
  3. Samples salient audio clips
    python get_salient_audioclips.py
    
  4. Samples salient frames
    python get_salient_frames.py
    
  5. Samples salient text
    python get_salient_text.py
    
  6. Now change back to project home
    cd ..
    
  7. Get glove matrix
    python get_glove_matrix.py
    
  8. Preprocess data
    python preprocess_text.py
    
  9. Prepare splits and vocab
    python prepare_data.py
    

Hyperparams

  • Hyper parameters are read from config.py

Train

python train.py