List your company

Broad-Domain Factual and Professional Reasoning on MMLU PRO

0.711Accuracy (ACC)Raw CoT
0.343880.439190.53450.62981Jun 28, 2026

Evaluation results

MethodLinks
Raw CoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.060.7111,789
CLSRLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.060.703356
SoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.060.692458
CoDLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.060.683582
CLSRLLM=QWEN3-32B2026.060.681118
Raw CoTLLM=QWEN3-32B2026.060.675405
CCoTLLM=QWEN3-32B2026.060.651192
CCoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.060.645516
SoTLLM=QWEN3-32B2026.060.643134
CoDLLM=QWEN3-32B2026.060.642318
CLSRLLM=QWEN3-8B2026.060.60496
Raw CoTLLM=QWEN3-8B2026.060.602276
SoTLLM=QWEN3-8B2026.060.585118
CoDLLM=QWEN3-8B2026.060.556125
CCoTLLM=QWEN3-8B2026.060.502135
CLSRLLM=LLaMA3-8B2026.060.389320
Raw CoTLLM=LLaMA3-8B2026.060.386960
CCoTLLM=LLaMA3-8B2026.060.372370
SoTLLM=LLaMA3-8B2026.060.364340
CoDLLM=LLaMA3-8B2026.060.358352