Implementation of Learning End-to-End Goal-Oriented Dialog with sklearn-like interface using Tensorflow. Tasks are from the bAbl dataset. Based on an earlier implementation (can't find the link).
pip install -r requirements.txt
python single_dialog.py
Train the model
python single_dialog.py --train True --task_id 1 --interactive False
Running a single bAbI task Demo
python single_dialog.py --train False --task_id 1 --interactive True
These files are also a good example of usage.
- tensorflow
- scikit-learn
- six
- scipy
Unless specified, the Adam optimizer was used.
The following params were used:
- epochs: 200
- learning_rate: 0.01
- epsilon: 1e-8
- embedding_size: 20
| Task | Training Accuracy | Validation Accuracy | Test Accuracy |
|---|---|---|---|
| 1 | 99.9 | 99.1 | 99.3 |
| 2 | 100 | 100 | 99.9 |
| 3 | 96.1 | 71.0 | 71.1 |
| 4 | 99.9 | 56.7 | 57.2 |
| 5 | 99.9 | 98.4 | 98.5 |
| 6 | 73.1 | 49.3 | 40.6 |