Share your thoughts, 1 month free Claude Pro on usSee more

Short-form generation on Short-form generation datasets 1D-DiffTask most OOD

0.64PRR

MSP

Updated 4d ago

Evaluation Results

Method	Links
MSP 2026.04		0.64
MSP 2026.04		0.57
SATRMD-MSP 2026.04		0.52
SATRMD 2026.04		0.34
SAPLMA (middle) 2026.04		0.32
SATMD-MSP 2026.04		0.27
HUQ-SATMD 2026.04		0.24
HUQ-SATRMD 2026.04		0.22
SATRMD-MSP 2026.04		0.21
SAPLMA (middle) 2026.04		0.21
Lookback-lens 2026.04		0.19
SAPLMA (top) 2026.04		0.18
HUQ-SATMD 2026.04		0.17
HUQ-SATRMD 2026.04		0.17
Uhead (1 layer) 2026.04		0.16
Uhead (2 layers) 2026.04		0.16
Lookback-lens 2026.04		0.16
SATMD 2026.04		0.15
SATRMD 2026.04		0.13
Uhead (1 layer) 2026.04		0.08
SATMD-MSP 2026.04		0.08
SAPLMA (top) 2026.04		0.08
SATMD 2026.04		0.02
Uhead (2 layers) 2026.04		-0.36