Alleviating Sequence Information Loss with Data Overlapping and Prime Batch Sizes

Open paper
Benchmark Model Rank Results
language-modelling-on-wikitext-103AWD-LSTM-MoS + ATOI#69Test perplexity: 32.85Validation perplexity: 31.92
language-modelling-on-wikitext-2AWD-LSTM + ATOI#29Test perplexity: 64.73Validation perplexity: 67.47…

Rank counts only results with a code link.