COMP4211 Assignment 2
Ng Chi Him SID 20420921 chngax@connect.ust.hk
Usage
The three tasks have been implemented on separate Jupyter Notebooks in Anaconda environment with Python 3.6.
To setup and activate the environment, run:
conda create -n comp4211 python=3.6 anaconda
conda install -n comp4211 -c pytorch pytorch torchvision
conda install -n comp4211 tensorflow tensorboardX
conda active comp4211
Then, launch Jupyter Notebook and open files using web ui:
jupyter notebook
Logs generated in the submission is stored in /logs
and can be viewed by tensorboard:
tensorboard --logdir ./logs/
Provided code and Dataset
The three provided files (pa2_sample_code.py
, pretrained_encoder.pt
and train_test_split.npz
) are stored in root directory of the project. They are included in the code submission but feel free to replace them with official copy when grading.
As instructed, the dataset was not included in code submission. To obtain the dataset, run pa2_sample_code.py
. (or place local copy to /data
when grading, folder structure is not modified.)