Skip to content

Latest commit

 

History

History
31 lines (18 loc) · 878 Bytes

File metadata and controls

31 lines (18 loc) · 878 Bytes

Setup

Config

Inside the config there are hardcoded paths to the data. (captions)

Despite Pro-Cap captions you need generic captions (there are also inside my folder /home2/faculty/mgalkowski/memes_analysis/data/hateful_memes/captions).

You also need data from hateful_memes/Pro-Cap/Data/mem*. I have put them inside /home2/faculty/mgalkowski/memes_analysis/data/hateful_memes/data/hateful_memes/domain_splits/. It is important that they are inside domain_splits folder.

Training

Before running training script you need to generate captions for the images.

Modify paths in generate_captions.py and run it.

sbatch generate_captions.sh

Modify those files before running training script

config.py - modify it to your needs.

train_model.sh - modify it to your needs.

Run training script

sbatch train_model.sh