List your company

Machine Reading Comprehension on QA-Feedback (test)

44.9RelevanceLCO-KLD
27.32431.88736.4541.013Mar 1, 2026

Evaluation results

MethodLinks
LCO-KLDBackbone=Mistral-3-3B2026.0344.967.36258.1
PPOBackbone=Mistral-3-3B2026.0343.161.453.152.5
SFTBackbone=Mistral-3-3B2026.0342.758.951.351
π*Backbone=Mistral-3-3B2026.0341.659.752.851.4
REINFORCEBackbone=Mistral-3-3B2026.0341.662.953.652.7
ϕDPOBackbone=Mistral-3-3B2026.0341.265.855.955.6
DAPOBackbone=Mistral-3-3B2026.0341.262.552.852.2
GRPOBackbone=Mistral-3-3B2026.034162.252.952
LCO-LCHBackbone=Mistral-3-3B2026.0339.663.957.753.7
LCO-MSEBackbone=Mistral-3-3B2026.0339.566.955.253.9
LCO-KLDBackbone=Llama-3.2-3B2026.0339.281.761.260.7
GSPOBackbone=Mistral-3-3B2026.0339.263.952.251.8
LCO-LCHBackbone=Llama-3.2-3B2026.0338.381.454.358
LCO-MSEBackbone=Llama-3.2-3B2026.0337.879.353.957
PPOBackbone=Llama-3.2-3B2026.0337.462.750.750.3
ϕDPOBackbone=Llama-3.2-3B2026.0336.973.754.955.2
SFTBackbone=Llama-3.2-3B2026.0335.259.748.847.9
GRPOBackbone=Llama-3.2-3B2026.0334.270.851.652.2
DAPOBackbone=Llama-3.2-3B2026.0333.772.151.652.5
GSPOBackbone=Llama-3.2-3B2026.0333.272.451.652.4
π*Backbone=Llama-3.2-3B2026.0332.169.151.450.9
REINFORCEBackbone=Llama-3.2-3B2026.032863.250.647.3