Releases: TharinduDR/Siamese-Recurrent-Architectures
Releases · TharinduDR/Siamese-Recurrent-Architectures
1.3.1
1.3.0
- Implemented Capsule network, shared capsule on top of GRU layer
- Implemented paragraph embeddings
- Implemented mergiong of different embeddings
1.2.0
- Added better documentation
- Added requirements file
- Added MALSTM-ATTENTION, MAGRU-ATTENTION modelsand compared it with the MALSTM.
1.1.0
- MABIGRU - Siamese recurrent network with Bi directional GRU was implemented. Performance increased compared to original paper - MALSTM, but not better than MAGRU in 1.0.0 release
- MABILSTM - Siamese recurrent network with Bi directional LSTM was implemented
1.0.0
- Implemented MALSTM
- Implemented different optimizers for MALSTM and compared it with the original approach - gained a performance improvement
- Implemented MAGRU with different optimizers and compared it to MALSTM - gained a performance improvement