List your company

Multi-hop Question Answering on HOTPOT-QA (ACC, TKN)

68.8AccuracyCLSR
39.05646.77854.562.222Jun 28, 2026

Evaluation results

MethodLinks
CLSRLLM=QWEN3-32B2026.0668.855
Raw CoTLLM=QWEN3-32B2026.0668.4165
Raw CoTLLM=QWEN3-8B2026.0666.4134
CLSRLLM=QWEN3-8B2026.0666.448
SoTLLM=QWEN3-32B2026.0666.272
SoTLLM=QWEN3-8B2026.0665.863
CCoTLLM=QWEN3-32B2026.0665.881
CCoTLLM=QWEN3-8B2026.0665.270
CoDLLM=QWEN3-32B2026.0664.678
CoDLLM=QWEN3-8B2026.0663.258
Raw CoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0655.8213
SoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0654.859
CLSRLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0654.543
CCoTLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.065365
CoDLLM=DEEPSEEK-R1-0528-QWEN3-8B2026.0651.256
SoTLLM=LLaMA3-8B2026.0645.821
CLSRLLM=LLaMA3-8B2026.0645.514
Raw CoTLLM=LLaMA3-8B2026.0645.132
CCoTLLM=LLaMA3-8B2026.0644.328
CoDLLM=LLaMA3-8B2026.0640.224