UL2 20B (fine-tuned)

Reported on 4 benchmarks across 3 tasks · 1 paper

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

Natural Language Processing4 results

Question AnsweringonCOPA
Accuracy· 2022-05-10
99
best: 100 (PaLM 540B (finetuned) )
UL2: Unifying Language Learning Paradigms arXiv:2205.05131
Question AnsweringonBoolQ
Accuracy· 2022-05-10
90.8
best: 99.87 (Mistral-Nemo 12B (HPT))
UL2: Unifying Language Learning Paradigms arXiv:2205.05131
Word Sense DisambiguationonWords in Context
Accuracy· 2022-05-10
77.3
best: 85.3 (COSINE + Transductive Learning)
UL2: Unifying Language Learning Paradigms arXiv:2205.05131
Coreference ResolutiononWinograd Schema Challenge
Accuracy· 2022-05-10
98.1
best: 100 (PaLM 540B (fine-tuned))
UL2: Unifying Language Learning Paradigms arXiv:2205.05131