Hint Token Learning Repository

Overview of the Project

This repository contains the data, training code, and best model parameters after hyper-parameter tuning in the hint token learning project.

Repository Structure

• 📂 Best_Model_Params/

  • KCNE1/: Parameters that obtain the best-performing models fine-tuned from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the KCNE1 dataset
  • PTEN/: Parameters that obtain the best-performing models fine-tuned from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the PTEN dataset
  • RecA/: Parameters that obtain the best-performing models fine-tuned from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the RecA dataset
  • TPMT/: Parameters that obtain the best-performing models fine-tuned from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the TPMT dataset

• 📂 Data/

  • KCNE1/: Original KCNE1 datasets used for training, validation, and testing
  • PTEN/: Original PTEN datasets used for training, validation, and testing
  • RecA/: Original RecA datasets used for training, validation, and testing
  • TPMT/: Original TPMT datasets used for training, validation, and testing

• 📂 Model_Training/

  • KCNE1/: Original training code to train models from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the KCNE1 dataset with and without the hint token learning strategy
  • PTEN/: Original training code to train models from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the PTEN dataset with and without the hint token learning strategy
  • RecA/: Original training code to train models from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the RecA dataset with and without the hint token learning strategy
  • TPMT/: Original training code to train models from ProtBERT, ESM_650M, ProtT5, and ESM_15B on the TPMT dataset with and without the hint token learning strategy

• 📄 environment.yml

  • Configuration file for environment settings