List your company

Question Answering on HotQA

48.5AccuracyCDKC
15.63624.16832.741.232Feb 13, 2026

Evaluation results

MethodLinks
CDKCBackbone=Qwen2.5-14B-Instruct2026.0248.5
GRPOBackbone=Qwen2.5-14B-Instruct2026.0248.17
CDKCBackbone=Qwen2.5-3B-Instruct2026.0240.07
GRPOBackbone=Qwen2.5-3B-Instruct2026.0238.67
RAGBackbone=Qwen2.5-14B-Instruct2026.0238.6
CGKEBackbone=Qwen2.5-14B-Instruct2026.0233.27
CoTBackbone=Qwen2.5-14B-Instruct2026.0232.9
RAGBackbone=Qwen2.5-3B-Instruct2026.0232.57
Vanilla SFTBackbone=Qwen2.5-14B-Instruct2026.0228
Vanilla LLMBackbone=Qwen2.5-14B-Instruct2026.0226.4
CGKEBackbone=Qwen2.5-3B-Instruct2026.0225.47
CoTBackbone=Qwen2.5-3B-Instruct2026.0221.43
Vanilla SFTBackbone=Qwen2.5-3B-Instruct2026.0219.9
Vanilla LLMBackbone=Qwen2.5-3B-Instruct2026.0216.9