GPT-NeoX 20B (few-shot, k=3)

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

Common Sense ReasoningonBIG-bench (Causal Judgment)
Accuracy· 2023-03-30
52.41
best: 62 (PaLM 2 (few-shot, k=3, Direct))
BloombergGPT: A Large Language Model for Finance arXiv:2303.17564
Common Sense ReasoningonBIG-bench (Disambiguation QA)
Accuracy· 2023-03-30
40.8
best: 78.8 (PaLM 2 (few-shot, k=3, Direct))
BloombergGPT: A Large Language Model for Finance arXiv:2303.17564
Common Sense ReasoningonBIG-bench (Date Understanding)
Accuracy· 2023-03-30
45.6
best: 91.2 (PaLM 2 (few-shot, k=3, CoT))
BloombergGPT: A Large Language Model for Finance arXiv:2303.17564

Methodology1 result

Logical ReasoningonBIG-bench (Formal Fallacies Syllogisms Negation)
Accuracy· 2023-03-30
52.8
best: 64.8 (PaLM 2 (few-shot, k=3, Direct))
BloombergGPT: A Large Language Model for Finance arXiv:2303.17564