TaLK Convolutions

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.

Natural Language Processing3 results

Machine TranslationonIWSLT2014 German-English
BLEU score· 2020-02-08
35.5
best: 40.43 (PiNMT)
Time-aware Large Kernel Convolutions arXiv:2002.03184
Machine TranslationonWMT2014 English-German
BLEU score· 2020-02-08
29.6
best: 35.14 (Transformer Cycle (Rev))
Time-aware Large Kernel Convolutions arXiv:2002.03184
Machine TranslationonWMT2014 English-French
BLEU score· 2020-02-08
43.2
best: 46.4 (Transformer+BT (ADMIN init))
Time-aware Large Kernel Convolutions arXiv:2002.03184

Medical1 result

Language ModellingonWikiText-103
Test perplexity· 2020-02-08
23.3
best: 2.4 (RETRO (7.5B))
Time-aware Large Kernel Convolutions arXiv:2002.03184