Students' names :
Amirahmad Shafiee
Ali Shafiei
Neda Fallah
This repo consists of two parts :
1. Document Classification:
Sentiment Analysis task
base-model : SVM on TF-IDF vectors
main-model : ParseBert
dataset : SentiPers(Persian dataset for multioutput polarity classification)
We benchmarked the model results and it happened to be 0.3 points of accuracy(on a scale of 1-100) behind the top benchmark.
2. Token Classification:
Question answering task
main-model :
base-model : bidirectional GRU
dataset :
Tags:
- BERT
- Transformers
- Generation
- Sentiment Analysis
- Question Answering
- Recurrent neural networks, bidirectional GRU
- Classic ML