AWD-LSTM-MoS + Partial Shuffle

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· 2019-03-11
53.92
best: 20.5 (GPT-3 (Zero-Shot))
Partially Shuffling the Training Data to Improve Language Models arXiv:1903.04167
Language ModellingonPenn Treebank (Word Level)
Validation perplexity· 2019-03-11
55.89
best: 36.1 (BERT-Large-CAS)
Partially Shuffling the Training Data to Improve Language Models arXiv:1903.04167
Language ModellingonWikiText-2
Test perplexity· 2019-03-11
59.98
best: 8.21 (SparseGPT (175B, 50% Sparsity))
Partially Shuffling the Training Data to Improve Language Models arXiv:1903.04167
Language ModellingonWikiText-2
Validation perplexity· 2019-03-11
62.38
best: 15.69 (GPT-2 (fine-tuned))
Partially Shuffling the Training Data to Improve Language Models arXiv:1903.04167