Our new X account is live! Follow @wizwand_team for updates

Commonsense Reasoning on WinoGrande

94.1Accuracy

Human Performance

Updated 2d ago

Evaluation Results

Method
Human Performance 2024.01	94.1	-	-	-	-	-	-
VERA-T5 (Multitask Supervised) 2024.01	92.4	-	-	-	-	-	-
PaLM 2 2023.05	90.9	-	-	-	-	-	-
BF16 2026.02	89.34	-	-	-	-	-	-
HiF4 2026.02	89.11	-	-	-	-0.23	-	-
NVFP4+PTS 2026.02	88	-	-	-	-1.34	-	-
NVFP4 2026.02	87.92	-	-	-	-1.42	-	-
Dense Model 2026.01	87.8	-	-	-	-	-	-
GPT-4 2023.05	87.5	-	-	-	-	-	-
Dense Model 2026.01	86.8	-	-	-	-	-	-
GradPruner 2026.01	86.1	-	-	-	-	-	-
GradPruner 2026.01	86	-	-	-	-	-	-
DeepSeek V3.1 2026.02	85.9	-	-	-	-	-	-
DeepSeek-V3.1 Base 2026.01	85.9	-	-	-	-	-	-
DeepSeek V3.2 2026.02	85.6	-	-	-	-	-	-
DeepSeek-V3.2 Exp Base 2026.01	85.6	-	-	-	-	-	-
Kimi-K2 2026.02	85.3	-	-	-	-	-	-
Kimi-K2 Base 2026.01	85.3	-	-	-	-	-	-
PaLM 2023.05	85.1	-	-	-	-	-	-
SE-GPT 2024.07	84.8	-	-	-	-	-	-
Dense Model 2026.01	84.6	-	-	-	-	-	-
DeBERTa-v3-L (Supervised) 2024.01	84.1	-	-	-	-	-	-
MiMo-V2 Flash 2026.02	83.8	-	-	-	-	-	-
MiMo-V2-Flash Base 2026.01	83.8	-	-	-	-	-	-
PaLM 2023.05	83.7	-	-	-	-	-	-
FT(Llama3.2) 2026.01	83.6	-	-	-	-	-	-
SAT 2026.01	83.6	-	-	-	-	-	-
SAT 2026.01	83.5	-	-	-	-	-	-
Zero-shot-CoT 2024.07	83.3	-	-	-	-	-	-
LLaMA 2 70B 2024.01	83.2	-	-	-	-	-	-
GradPruner 2026.01	83.2	-	-	-	-	-	-
PaLM 2-L 2023.05	83	-	-	-	-	-	-
MINITRON 2026.01	83	-	-	-	-	-	-
Self-EXP 2024.07	82.8	-	-	-	-	-	-
Laco 2026.01	82.8	-	-	-	-	-	-
Self-ICL-CoT 2024.07	82.6	-	-	-	-	-	-
Dense Model 2026.01	82.1	-	-	-	-	-	-
Laco 2026.01	81.8	-	-	-	-	-	-
Mistral Nemo Inst 2407 2026.01	81.7	-	-	-	-	7.2	-
GPT-3.5 2024.01	81.6	-	-	-	-	-	-
LLMPruner 2026.01	81.5	-	-	-	-	-	-
SAT 2026.01	81.5	-	-	-	-	-	-
APT 2026.01	81.5	-	-	-	-	-	-
MINITRON 2026.01	81.4	-	-	-	-	-	-
Mixtral 8x7B 2024.01	81.2	-	-	-	-	-	-
Zero-shot 2024.07	81.2	-	-	-	-	-	-
GradPruner 2026.01	81.2	-	-	-	-	-	-
Laco 2026.01	81.2	-	-	-	-	-	-
Qwen3 0.6B 2026.01	81.2	-	-	-	-	28.3	-
Qwen3 0.6B 2026.01	81.2	-	-	-	-	-	-
PaLM-540B 2023.09	81.1	-	-	-	-	-	-
LLMPruner 2026.01	81	-	-	-	-	-	-
Llama-3-70B 2025.03	80.9	-	-	-	-	-	-
FT(Llama3.2) 2026.01	80.8	-	-	-	-	-	-
MINITRON 2026.01	80.5	-	-	-	-	-	-
LO-BCQ (g128) 2025.02	80.43	-	-	-	-	-	-
LLaMA 2 70B 2024.01	80.4	-	-	-	-	-	-
BF16 2026.02	80.27	-	-	-	-	-	-
LLMPruner 2026.01	80.1	-	-	-	-	-	-
LO-BCQ (g128) 2025.02	79.79	-	-	-	-	-	-
F-L-S 2024.10	79.72	-	-	-	-	-	-
Mistral Small 24B Inst 2501 2026.01	79.7	-	-	-	-	3.5	-
HiF4 2026.02	79.56	-	-	-	-0.71	-	-
Self-ICL 2024.07	79.5	-	-	-	-	-	-
NVFP4+PTS 2026.02	79.4	-	-	-	-0.87	-	-
ROBERTa-L (Supervised) 2024.01	79.3	-	-	-	-	-	-
NVFP4 2026.02	79.24	-	-	-	-1.03	-	-
LO-BCQ (g128) 2025.02	79.24	-	-	-	-	-	-
PaLM 2-M 2023.05	79.2	-	-	-	-	-	-
Laco 2026.01	79.2	-	-	-	-	-	-
Step 3.5 Flash 2026.02	79.1	-	-	-	-	-	-
Best Single Model 2024.10	79.01	-	-	-	-	-	-
LLMPruner 2026.01	79	-	-	-	-	-	-
APT 2026.01	79	-	-	-	-	-	-
Model-GLUE 2024.10	78.85	-	-	-	-	-	-
GRP-Oblit-1 2026.02	78.8	-	-	-	-	-	-
Abliteration 2026.02	78.6	-	-	-	-	-	-
Llama 3.1 8B 2026.01	78.5	-	-	-	-	8	-
Mistral 7B v0.3 2026.01	78.5	-	-	-	-	6.9	-
Mistral Nemo Base 2407 2026.01	78.5	-	-	-	-	4.8	-
Llama 3.1 8B 2026.01	78.5	-	-	-	-	8	-
Mistral 7B v0.3 2026.01	78.5	-	-	-	-	6.9	-
Mistral Nemo Base 2407 2026.01	78.5	-	-	-	-	4.8	-
DARE 2024.10	78.37	-	-	-	-	-	-
DeBERTa-v3-L (CANDLE Distilled) 2024.01	78.3	-	-	-	-	-	-
LO-BCQ (g128) 2025.02	78.3	-	-	-	-	-	-
CAR-DEBERTa-v3-L 2024.01	78.2	-	-	-	-	-	-
Abliteration 2026.02	78.2	-	-	-	-	-	-
CAR-DEBERTa-v3-L 2024.01	78.1	-	-	-	-	-	-
MINITRON 2026.01	78.1	-	-	-	-	-	-
Base 2026.02	78.1	-	-	-	-	-	-
SAT 2026.01	78	-	-	-	-	-	-
GRP-Oblit-1 2026.02	78	-	-	-	-	-	-
GRP-Oblit-1 2026.02	78	-	-	-	-	-	-
PaLM 2-S 2023.05	77.9	-	-	-	-	-	-
TIES 2024.10	77.9	-	-	-	-	-	-
GRP-Oblit 2026.02	77.9	-	-	-	-	-	-
GRP-Oblit 2026.02	77.9	-	-	-	-	-	-
GRP-Oblit-1 2026.02	77.7	-	-	-	-	-	-
Base 2026.02	77.6	-	-	-	-	-	-

Showing 100 of 792 rows

...