NLP-Challenge-Polynomial-Expansion

Implement a deep neural network model that learns to expand single variable polynomials. Model input is factorized sequence and output is predicted expanded sequence.

Problem Details

Report

Example

(7-3z)(-5z-9)=15z**2-8*z-63

(7-3z)(-5*z-9) is the factorized form

15z**2-8z-63 is the expanded form

Dataset and model

Dataset: dataset.txt

Trained model: seq2seq.h5

Dataset statistics: data_summary.ipynb

Reproduce

1. Proprocess the dataset and split it into train.txt, valid.txt and test.txt.(use --help for more options):

   python preprocess.py --dataset_path dataset.txt --seed 2022
   

2. Train the Seq2Seq model on train.txt and valid.txt:

   python train.py --hidden_dim 256 --batch_size 1024 --epochs 10 --learning_rate 0.005 --workers 2
   

3. Test the Seq2Seq model on test.txt:

   python test.py --batch_size 1024 --model_path seq2seq.h5 --test_path test.txt --workers 2
   

Performance

Accuracy and loss on test data:

Test accuracy: 0.9859623312950134

Test loss: 0.038746245205402374

Training and Validation Accuracy curve:

Training and Validation Loss curve:

References

https://blog.keras.io/a-ten-minute-introduction-to-sequence-to-sequence-learning-in-keras.html