Neo-6B (few-shot)

Reported on 5 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 Processing5 results

Question AnsweringonCOPA
Accuracy· 2022-10-05
77
best: 100 (PaLM 540B (finetuned) )
Ask Me Anything: A simple strategy for prompting language models arXiv:2210.02441
Question AnsweringonStory Cloze
Accuracy· 2022-10-05
51
best: 87.8 (Neo-6B (QA + WS))
Ask Me Anything: A simple strategy for prompting language models arXiv:2210.02441
Question AnsweringonMultiRC
F1· 2022-10-05
60.8
best: 90.1 (PaLM 540B (finetuned) )
Ask Me Anything: A simple strategy for prompting language models arXiv:2210.02441
Question AnsweringonBoolQ
Accuracy· 2022-10-05
66.5
best: 99.87 (Mistral-Nemo 12B (HPT))
Ask Me Anything: A simple strategy for prompting language models arXiv:2210.02441
Coreference ResolutiononWinograd Schema Challenge
Accuracy· 2022-10-05
36.5
best: 100 (PaLM 540B (fine-tuned))
Ask Me Anything: A simple strategy for prompting language models arXiv:2210.02441