List your company

Long-form generation on DiffTask OOD

0.16PRRHUQ-SATRMD
-0.1936-0.1018-0.010.0818Apr 13, 2026

Evaluation results

MethodLinks
HUQ-SATRMDModel Backbone=Llama 3.1-8B2026.040.16
SATMD-MSPModel Backbone=Llama 3.1-8B2026.040.15
SAPLMA (middle)Model Backbone=Llama 3.1-8B2026.040.1
SATRMD-MSPModel Backbone=Llama 3.1-8B2026.040.09
SAPLMA (top)Model Backbone=Llama 3.1-8B2026.040.05
Uhead (1 layer)Model Backbone=Gemma-2-9B2026.040.05
Uhead (2 layers)Model Backbone=Gemma-2-9B2026.040.05
SATMDModel Backbone=Llama 3.1-8B2026.040.04
HUQ-SATMDModel Backbone=Llama 3.1-8B2026.040.04
SATRMDModel Backbone=Llama 3.1-8B2026.040.03
SAPLMA (middle)Model Backbone=Gemma-2-9B2026.040.02
HUQ-SATMDModel Backbone=Gemma-2-9B2026.040.02
MSPModel Backbone=Llama 3.1-8B2026.04-0.01
SAPLMA (top)Model Backbone=Gemma-2-9B2026.04-0.01
Lookback-lensModel Backbone=Llama 3.1-8B2026.04-0.02
SATMDModel Backbone=Gemma-2-9B2026.04-0.02
Uhead (2 layers)Model Backbone=Llama 3.1-8B2026.04-0.03
SATRMD-MSPModel Backbone=Gemma-2-9B2026.04-0.06
Uhead (1 layer)Model Backbone=Llama 3.1-8B2026.04-0.07
Lookback-lensModel Backbone=Gemma-2-9B2026.04-0.07
MSPModel Backbone=Gemma-2-9B2026.04-0.1
HUQ-SATRMDModel Backbone=Gemma-2-9B2026.04-0.11
SATRMDModel Backbone=Gemma-2-9B2026.04-0.12
SATMD-MSPModel Backbone=Gemma-2-9B2026.04-0.18