Seq-U-Net

Reported on 4 benchmarks across 2 tasks · 1 paper

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

Medical2 results

Language ModellingonPenn Treebank (Word Level)
Test perplexity· 2019-11-14
107.95
best: 20.5 (GPT-3 (Zero-Shot))
Seq-U-Net: A One-Dimensional Causal U-Net for Efficient Sequence Modelling arXiv:1911.06393
Language ModellingonPenn Treebank (Character Level)
Bit per Character (BPC)· 2019-11-14
1.3
best: 1.38 (Bipartite Flow)
Seq-U-Net: A One-Dimensional Causal U-Net for Efficient Sequence Modelling arXiv:1911.06393

Music2 results

Music ModelingonNottingham
NLL· 2019-11-14
2.97
best: 4.05 (RNN)
Seq-U-Net: A One-Dimensional Causal U-Net for Efficient Sequence Modelling arXiv:1911.06393
Music ModelingonJSB Chorales
NLL· 2019-11-14
8.173
best: 8.54 (GRU)
Seq-U-Net: A One-Dimensional Causal U-Net for Efficient Sequence Modelling arXiv:1911.06393