List your company

Question Answering on HotPotQA (test) (Custom Metrics: Acc, Prec, Abs(0), Delta)

25.52AccuracyAFH (Absolute)
-1.02085.869612.7619.6504Oct 1, 2025

Evaluation results

MethodLinks
AFH (Absolute)Backbone=Qwen2.5-7B-Base2025.1025.5236.7340.327.93
DPOBackbone=Qwen2.5-7B-Base2025.1024.3548.2572.860.4
5-shot PromptingBackbone=Qwen2.5-7B-Base2025.1023.0634.1739.4524.18
MASH w/ OTC-STBackbone=Qwen2.5-7B-Base2025.1020.9860.6787.1464.51
MASH w/ OTC-STBackbone=Qwen3-4B-Base2025.1020.9653.5981.6367.07
AFH (Multisample)Backbone=Qwen2.5-7B-Base2025.1017.6851.368253.22
5-shot PromptingBackbone=Qwen3-4B-Base2025.1017.5834.3759.7643.72
DPOBackbone=Qwen3-4B-Base2025.1016.7260.8492.3875.87
AFH (Absolute)Backbone=Qwen3-4B-Base2025.1013.4447.9782.8141.5
AFH (Multisample)Backbone=Qwen3-4B-Base2025.107.2566.5496.5436.51
OTCBackbone=Qwen2.5-7B-Base2025.100—1000
OTCBackbone=Qwen3-4B-Base2025.100—1000