List your company

Question Answering on NQ-Open

47.4Exact Match (EM)MD
-1.89610.90223.736.498Mar 5, 2025May 25, 2025Aug 15, 2025Nov 4, 2025Jan 25, 2026Apr 16, 2026Jul 7, 2026

Evaluation results

MethodLinks
MDModel=Llama 32025.0347.4
PTRModel=GPT-5.42026.0446.9
MARGOBackbone=Qwen3-8B2026.0742.24
FixedThink+RLBackbone=Qwen3-8B2026.0740.66
MDModel=Llama 22025.0340.5
AdaptiveRLBackbone=Qwen3-8B2026.0740.08
FixedThinkBackbone=Qwen3-8B2026.0739.94
GreedyModel=Llama 32025.0339.6
FixedNoThink+RLBackbone=Qwen3-8B2026.0739.09
FSCModel=Llama 32025.0338.7
FixedNoThinkBackbone=Qwen3-8B2026.0738.64
USCModel=Llama 22025.0338.6
BoNModel=Llama 32025.0338.6
IDModel=Llama 32025.0338.5
AdaptiveCLSBackbone=Qwen3-8B2026.0738.5
AdaptiveSFTBackbone=Qwen3-8B2026.0738.5
PTRModel=Claude Sonnet 4.62026.0438
SRModel=Llama 22025.0337.3
USCModel=Llama 32025.0337
SRModel=Llama 32025.0336.9
GreedyModel=Llama 22025.0336.6
DoLaModel=Llama 32025.0335.7
BoNModel=Llama 22025.0335.2
MARGOBackbone=Qwen3-4B2026.0734.76
FixedThink+RLBackbone=Qwen3-4B2026.0733.8
DoLaModel=Llama 22025.0333.7
FSCModel=Llama 22025.0333.5
IDModel=Llama 22025.0333.1
FixedThinkBackbone=Qwen3-4B2026.0732.94
AdaptiveRLBackbone=Qwen3-4B2026.0732.49
MDModel=Gemma 22025.0331
FixedNoThink+RLBackbone=Qwen3-4B2026.0730.61
AdaptiveCLSBackbone=Qwen3-4B2026.0730.33
AdaptiveSFTBackbone=Qwen3-4B2026.0730.28
PTRModel=Claude Haiku 4.52026.0430
FixedNoThinkBackbone=Qwen3-4B2026.0729.5
SRModel=Gemma 22025.0328
PTRModel=GPT-4o-Mini2026.0426
FSCModel=Gemma 22025.0324
USCModel=Gemma 22025.0323.9
GreedyModel=Gemma 22025.0323
BoNModel=Gemma 22025.0319
DoLaModel=Gemma 22025.0318.2
IDModel=Gemma 22025.0316.5
ReActModel=GPT-5.42026.0416
ReActModel=GPT-4o-Mini2026.046
ReActModel=Claude Haiku 4.52026.042
ReActModel=Claude Sonnet 4.62026.040