ML Model Evaluation with YJMob100K

Organization

|-- Analysis.ipynb                      # Some Data Analysis Done
|-- Baseline.py                         # A Visit Frequency-based Predictive Model
|-- CustomTransformer_with_YJMob100K.py # Custom Transformer
|-- LSTM_with_YJmob100K.py              # Built-in LSTM
|-- README.md
|-- Transformer_with_YJMob100K.py       # Built-in Transformer
|-- Visualization.ipynb                 # Image visualizaitons for final presentation
|-- input_data_processing.ipynb         # Notebook file for Data Processing

Run Baseline

python Baseline.py

Environment Set-Up (for Running LSTM and Transformer)

1. Sign into NYU HPC:

https://sites.google.com/nyu.edu/nyu-hpc/accessing-hpc

2. Clone Repository

git clone https://github.com/ANNIZHENG/MLModel_Eval_with_YJMob100K.git

3. Request resourses (2 hour, 4 cores, 1 GPU)

srun -t 2:00:00 -c 4 --mem=16000 --gres=gpu:1 --pty /bin/bash

4. Check CUDA and Python Availability (Optional)

module spider cuda
module spider python

5. Load necessary packages

module load cuda/11.6.2
module load python/intel/3.8.6

6. Set up Virtual Environment

virtualenv --system-site-packages -p python3 ./venv
pip install torch

7. Activate Virtual Environment

source ./venv/bin/activate

8. Run Files

python <FILE_NAME>.py