One For All & All For One: Bypassing Hyperparameter Tuning with Model Averaging For Cross-Lingual Transfer.
Fabian David SchmidtIvan VulicGoran GlavasPublished in: CoRR (2023)
Keyphrases
- cross lingual
- model averaging
- transfer learning
- posterior distribution
- machine translation
- hyperparameters
- bayesian methods
- gaussian processes
- language modeling
- model selection
- supervised classification
- text classification
- parameter settings
- language model
- semi supervised learning
- text categorization
- cross validation
- closed form
- document clustering
- news articles
- gaussian process
- reinforcement learning
- information retrieval systems
- active learning
- high dimensional
- parameter estimation
- markov networks
- em algorithm
- support vector