Download presentation
Presentation is loading. Please wait.
Published byMichael Bro Modified over 5 years ago
2
Baseline Model CSV Files Pandas DataFrame Sentence Lists
Average Pooling Layer Sigmoid Output Layer Baseline Model Array Dataset Dataloader to Iterator Encoder (from list of vectors to vector) Embedding Layer (word2vec) Two layer of LSTM (pretrained LM)
3
Pretrain (Glove/BERT) and fine-tuning
Possible improvements for baseline For language model Trained on new corpus Glove fine-tuning BERT fine-tuning For network model Gated recurrent unit Attention mechanism For output Threshold Questions (Corpus) Encoder Output (0/1) 01 03 05 02 04 Pretrain (Glove/BERT) and fine-tuning Neural network (GRU, attention)
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.