Feed-Forward Blocks Control Contextualization in Masked Language Models.
Goro KobayashiTatsuki KuribayashiSho YokoiKentaro InuiPublished in: CoRR (2023)
Keyphrases
- language model
- feed forward
- language modeling
- back propagation
- document retrieval
- n gram
- artificial neural networks
- neural network
- neural nets
- probabilistic model
- language modelling
- speech recognition
- test collection
- query expansion
- hidden layer
- statistical language models
- information retrieval
- ad hoc information retrieval
- control system
- query terms
- pwm rectifier
- smoothing methods
- retrieval model
- document ranking
- feed forward neural networks
- vector space model
- context sensitive
- bayesian networks
- pseudo relevance feedback
- relevance model
- term dependencies
- document length
- error rate
- translation model
- expert systems
- decision making
- learning algorithm
- language models for information retrieval
- cross lingual