ChartDialog

Dataset download

https://drive.google.com/file/d/1B8tKdmI3LlAMsPQQW6KsCbTKfsm9vnFA/view?usp=sharing

How to run the baseline (seq2seq)

Dialog preprocessing

This step transforms dialogs into seq2seq-style files (input, gold output, etc.)

bash dialog_preprocess.sh [YOUR PATH]/ChartDialog-data [YOUR PATH]/ChartDialog-save

[YOUR PATH] is the directory under which you put the ChartDialog-data folder.

Model-specific preprocessing

cd models/seq2seq
bash preprocess.sh [YOUR PATH]/ChartDialog-save/s2s_single

The granularity single is used in this example; it can also be pair or split. For more information about the granularity please refer to our paper.

Training

bash train.sh [YOUR PATH]/ChartDialog-save/s2s_single [YOUR PATH]/ChartDialog-save/save_models s2s_single.1

The parameter setting in train.sh was used in our experiments in the paper.

Translating (Generating output)

bash translate.sh [YOUR PATH]/ChartDialog-save/s2s_single [YOUR PATH]/ChartDialog-save/save_models s2s_single.1 100000

Evaluating

cd ../..
bash evaluate.sh [YOUR PATH]/ChartDialog-save/s2s_single ./models/seq2seq/outputs/pred_test_s2s_single.1_100000.txt single

Pretrained model download

• seq2seq: https://drive.google.com/open?id=15Pv1IeAas0L1PAqCaDPOqiBvKhFonGft

Plotting Agent Interface

To run the interface, set up the environment by:

pip install -r requirements.txt

The plotting agent interface is inside models/seq2seq/Plotting Agent.ipynb