Improving Stability of Fine-Tuning Pretrained Language Models via Component-Wise Gradient Norm Clipping.
Chenghao YangXuezhe MaPublished in: CoRR (2022)
Keyphrases
- language model
- fine tuning
- language modeling
- probabilistic model
- document retrieval
- n gram
- speech recognition
- viable alternative
- fine tune
- retrieval model
- query expansion
- information retrieval
- statistical language models
- test collection
- language modelling
- context sensitive
- ad hoc information retrieval
- language models for information retrieval
- fine tuned
- relevance model
- vector space model
- translation model
- passage retrieval
- document ranking
- smoothing methods
- cross lingual
- query terms
- language model for information retrieval