Repo for Master's Dissertation and working paper: "Multimodal Prompt-Tuning with Human Cognition Data" by Yingjia Wan.
The implementation code will be fully uploaded upon conference submission.
-
The original ZuCo corpus is available at https://osf.io/uxamg/. As the data files were large, I recommend downloading them efficiently using my code at
./zuco_preprocessing/src/zuco_matfiles_download.py
. See systematic instructions below. -
This project also requires a
zuco_data_storage
folder inside the current dir, which is imported from https://github.com/DS3Lab/zuco-nlp/tree/master by Nora Hollenstein. I rename the folder namezuco-nlp
tozuco_data_storage
. Please refer to the README.md in thezuco_data_storage
folder for details.
Please refer to the subsequent README.md files in each folder for the instructions of setting up the environment for each step.
First, please refer to the README.md in the folder zuco_preprocessing
for data formatting and preprocessing instructions.
Please refer to the README.md in the folder AddToken_Method
for the instructions of Experiment 1.
Please refer to the README.md in the folder COGMAP
for the instructions of Experiment 2.