List your company

Question Answering on SQuAD (Accuracy, CR, OR)

99.6AccuracyLayer-LR
22.22442.31262.482.488Feb 26, 2026Mar 10, 2026Mar 23, 2026Apr 4, 2026Apr 17, 2026Apr 29, 2026May 12, 2026

Evaluation results

MethodLinks
Layer-LRModel=Qwen2.5-7B, Mode=FTG (First-token generation), Context=Out-of-domain2026.0299.6——
Layer-LRModel=Llama-3.1-8B, Mode=FTG (First-token generation), Context=Out-of-domain2026.0297.7——
Layer-LRModel=Mistral-7B-v0.1, Mode=FTG (First-token generation), Context=Out-of-domain2026.0291.8——
DCRDModel=Llama3-8B2026.0588.9——
GreedyModel=Llama3-8B2026.0587.2——
ADACADModel=Llama3-8B2026.0587.1——
COIECDModel=Llama3-8B2026.0586.9——
DCRDModel=Llama2-7B2026.0583.2——
DCRDModel=Llama2-13B2026.0579.5——
COIECDModel=Llama2-13B2026.0578.9——
ADACADModel=Llama2-13B2026.0576.5——
COIECDModel=Llama2-7B2026.0576——
DGAOBackbone=Llama-3.2-3B2026.0575.4489.617.57
Pos-BeginningBackbone=Llama-3.2-3B2026.0575.2192.3515.85
ADACADModel=Mistral-7B2026.0575.2——
Pos-EndBackbone=Llama-3.2-3B2026.0574.9990.4917.65
ADACADModel=Llama2-7B2026.0574.6——
DEmOBackbone=Llama-3.2-3B2026.0574.3489.2518.25
GreedyModel=Llama2-13B2026.0573.9——
GlobalEBackbone=Llama-3.2-3B2026.0573.8189.1518.3
SFTBackbone=Llama-3.2-3B2026.0573.7388.9818.54
SPHSBackbone=Llama-3.2-3B2026.0573.5989.118.71
PromptBackbone=Llama-3.2-3B2026.0573.3789.1418.9
CADModel=Llama2-13B2026.0572.7——
CADModel=Llama3-8B2026.0572.6——
GreedyModel=Llama2-7B2026.0571.6——
DCRDModel=Mistral-7B2026.0569.1——
CADModel=Llama2-7B2026.0567.7——
GreedyModel=Mistral-7B2026.0567.5——
PINEBackbone=Llama-3.2-3B2026.0566.9181.7518.42
COIECDModel=Mistral-7B2026.0548.1——
CADModel=Mistral-7B2026.0525.2——