Physical Commonsense Reasoning on PIQA (PIQA Score)
81.1PIQA ScoreLLAMA-4-SCOUT
Evaluation Results
| Method | Links | |
|---|---|---|
| LLAMA-4-SCOUTPARAMS=109B, Variant=ORIGINAL2025.11 | 81.1 | |
| LLAMA-4-SCOUTPARAMS=109B, Variant=FCSD2025.11 | 80.8 | |
| LLAMA-4-SCOUTPARAMS=109B, Variant=SFT2025.11 | 80.7 | |
| QWEN-3-30B MOEPARAMS=30B, Variant=ORIGINAL2025.11 | 80.5 | |
| QWEN-3-30B MOEPARAMS=30B, Variant=FCSD2025.11 | 80.4 | |
| DEEPSEEK-V2-LITEPARAMS=16B, Variant=ORIGINAL2025.11 | 80.1 | |
| DEEPSEEK-V2-LITEPARAMS=16B, Variant=FCSD2025.11 | 79.9 | |
| DEEPSEEK-V2-LITEPARAMS=16B, Variant=SFT2025.11 | 78.2 | |
| QWEN-3-30B MOEPARAMS=30B, Variant=SFT2025.11 | 77.8 | |
| Echo-180MParams=180M, Tokens=10B, Zero-shot=true2026.05 | 70.9 | |
| Transformer-180MParams=180M, Tokens=100B, Zero-shot=true2026.05 | 67.1 | |
| Mamba-2-180MParams=180M, Tokens=100B, Zero-shot=true2026.05 | 66.8 | |
| Mamba-3-MIMO-180MParams=180M, Tokens=100B, Zero-shot=true2026.05 | 66.7 | |
| GDN-180MParams=180M, Tokens=100B, Zero-shot=true2026.05 | 66.3 | |
| Mamba-3-SISO-180MParams=180M, Tokens=100B, Zero-shot=true2026.05 | 66.1 | |
| Echo-50MParams=50M, Tokens=3B, Zero-shot=true2026.05 | 59.5 |