UL2 20B (chain-of-thought)
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 Processing3 results
- Accuracy· 2022-05-1042.9best: 96.4 (GPT-4 (few-shot, k=25))
- Accuracy· 2022-05-1038.4best: 95.2 (ST-MoE-32B 269B (fine-tuned))
- Accuracy· 2022-05-1051.4best: 92.54 (GPT-4o (HPT))
Reasoning2 results
- Accuracy· 2022-05-104.4best: 97.72 (Claude 3.5 Sonnet (HPT))
- Parameters (Billion)· 2022-05-1020best: 540 (PaLM 540B (Self Improvement, Self Consistency))