Reward Modeling on TL;DR Seen (n=100) benchmark leaderboard · SOTA2 Research

Reward Modeling on TL;DR Seen (n=100)

62.3AccuracyVRF

Evaluation Results

Method	Links
VRF2026.04		62.3
VRF2026.04		61.9
LoRe2026.04		61.6
LoRe2026.04		61.3
PAL2026.04		60.4
BT2026.04		59.9
PAL2026.04		59.8
BT2026.04		59.5
VPL2026.04		58.6
VPL2026.04		58.1
PReF2026.04		56.2
PReF2026.04		55.7
Ref2026.04		51.9
Ref2026.04		51.8