Multi-dimensional Hierarchical Temporal Alignment for Improved Temporal Commonsense Reasoning in Large Language Models

Setup Instructions

To run this project, please follow the steps below:

1. Install Requirements

First, install the necessary dependencies for the project by running:

pip install -r requirements.txt

2. Download The Test Dataset

Go to MCTACO and download the dataset. Place the downloaded dev data in the data folder at the project root.

3. Configure

Open the configs/config.yaml file and modify the test field to point to the path of the dev data you just downloaded. For example:

test: "data/your_dataset_folder"

4. Set Model Name

In conf/config.yaml, locate the base_model_name field and set the name of the model you want to use. For example:

base_model_name: "mistralai/Mistral-7B-Instruct-v0.1"

5. Run the Main Program

Once the above steps are complete, you can run the main program by executing:

python main.py
# MHTA