FLAN 137B (prompt-tuned)

Reported on 5 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 Processing5 results

Question AnsweringonCOPA
Accuracy· 2021-09-03
94
best: 100 (PaLM 540B (finetuned) )
Finetuned Language Models Are Zero-Shot Learners arXiv:2109.01652
Question AnsweringonMultiRC
F1· 2021-09-03
83.4
best: 90.1 (PaLM 540B (finetuned) )
Finetuned Language Models Are Zero-Shot Learners arXiv:2109.01652
Question AnsweringonBoolQ
Accuracy· 2021-09-03
86.3
best: 99.87 (Mistral-Nemo 12B (HPT))
Finetuned Language Models Are Zero-Shot Learners arXiv:2109.01652
Common Sense ReasoningonReCoRD
EM· 2021-09-03
85.1
best: 95.9 (Turing NLR v5 XXL 5.4B (fine-tuned))
Finetuned Language Models Are Zero-Shot Learners arXiv:2109.01652
Coreference ResolutiononWinograd Schema Challenge
Accuracy· 2021-09-03
86.5
best: 100 (PaLM 540B (fine-tuned))
Finetuned Language Models Are Zero-Shot Learners arXiv:2109.01652