Head2toe: Utilizing intermediate representations for better transfer learning

U Evci, V Dumoulin, H Larochelle… - … on Machine Learning, 2022 - proceedings.mlr.press
Transfer-learning methods aim to improve performance in a data-scarce target domain using
a model pretrained on a data-rich source domain. A cost-efficient strategy, linear probing …
被引用次数：72 相关文章

Head2Toe: Utilizing Intermediate Representations for Better Transfer Learning

U Evci, V Dumoulin, H Larochelle, MC Mozer - arXiv e-prints, 2022 - ui.adsabs.harvard.edu
Transfer-learning methods aim to improve performance in a data-scarce target domain using
a model pretrained on a data-rich source domain. A cost-efficient strategy, linear probing …

Head2Toe: Utilizing Intermediate Representations for Better Transfer Learning

U Evci, V Dumoulin, H Larochelle, M Mozer - research.google
Transfer-learning methods aim to improve performance in a data-scarce target domain using
a model pretrained on a source domain. A cost-efficient strategy,, involves freezing the …

Head2Toe: Utilizing Intermediate Representations for Better Transfer Learning

U Evci, V Dumoulin, H Larochelle… - arXiv preprint arXiv …, 2022 - arxiv.org
Transfer-learning methods aim to improve performance in a data-scarce target domain using
a model pretrained on a data-rich source domain. A cost-efficient strategy, linear probing …

Head2Toe: Utilizing Intermediate Representations for Better Transfer Learning

U Evci, V Dumoulin, H Larochelle, M Mozer - research.google
Transfer-learning methods aim to improve performance in a data-scarce target domain using
a model pretrained on a source domain. A cost-efficient strategy,, involves freezing the …