AWD-LSTM 3-layer with Fraternal dropout

Reported on 4 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Medical4 results

Language ModellingonPenn Treebank (Word Level)
Test perplexity· 2017-10-31
56.8
best: 20.5 (GPT-3 (Zero-Shot))
Fraternal Dropout arXiv:1711.00066
Language ModellingonPenn Treebank (Word Level)
Validation perplexity· 2017-10-31
58.9
best: 36.1 (BERT-Large-CAS)
Fraternal Dropout arXiv:1711.00066
Language ModellingonWikiText-2
Test perplexity· 2017-10-31
64.1
best: 8.21 (SparseGPT (175B, 50% Sparsity))
Fraternal Dropout arXiv:1711.00066
Language ModellingonWikiText-2
Validation perplexity· 2017-10-31
66.8
best: 15.69 (GPT-2 (fine-tuned))
Fraternal Dropout arXiv:1711.00066