Symmetric Dot-Product Attention for Efficient Training of BERT Language Models.
Martin CourtoisMalte OstendorffLeonhard HennigGeorg RehmPublished in: ACL (Findings) (2024)
Keyphrases
- language model
- language modeling
- probabilistic model
- dot product
- document retrieval
- n gram
- statistical language models
- language modelling
- speech recognition
- information retrieval
- retrieval model
- test collection
- mixture model
- query expansion
- kernel function
- kernel methods
- smoothing methods
- image features
- training set
- bayesian networks