List your company

Ranking on UltraFeedback clean (held-out)

0.362Kendall’s TauNominal PL
-0.072720.040140.1530.26586Jul 2, 2026

Evaluation results

MethodLinks
Nominal PLModel=Qwen3-8B, Noise condition=Clean2026.070.362
Robust PLModel=Qwen3-8B, Noise condition=near_tie, epsilon=0.4, rho=0.052026.070.356
Nominal PLModel=Qwen3-8B, Noise condition=near_tie, epsilon=0.42026.070.348
Robust PLModel=Qwen3-8B, Noise condition=Clean, rho=0.052026.070.347
Robust PLModel=Qwen3-8B, Noise condition=Clean, rho=0.102026.070.34
TV-DR-DPOModel=Qwen3-8B, Noise condition=Clean, rho=0.102026.070.338
Robust PLModel=Qwen3-8B, Noise condition=near_tie, epsilon=0.4, rho=0.102026.070.333
KLDPOModel=Qwen3-8B, Noise condition=Clean, tau=1.002026.070.331
Robust PLModel=Qwen3-8B, Noise condition=top_rank, epsilon=0.4, rho=0.052026.070.329
Robust PLModel=Qwen3-8B, Noise condition=near_tie, epsilon=1.0, rho=0.052026.070.327
Nominal PLModel=Qwen3-8B, Noise condition=near_tie, epsilon=1.02026.070.322
Nominal PLModel=Qwen3-8B, Noise condition=top_rank, epsilon=0.42026.070.321
TV-DR-DPOModel=Qwen3-8B, Noise condition=near_tie, epsilon=0.4, rho=0.102026.070.319
Robust PLModel=Qwen3-8B, Noise condition=near_tie, epsilon=1.0, rho=0.102026.070.319
KLDPOModel=Qwen3-8B, Noise condition=near_tie, epsilon=0.4, tau=1.002026.070.318
Nominal BT (DPO)Model=Qwen3-8B, Noise condition=Clean2026.070.316
TV-DR-DPOModel=Qwen3-8B, Noise condition=near_tie, epsilon=1.0, rho=0.102026.070.31
KLDPOModel=Qwen3-8B, Noise condition=near_tie, epsilon=1.0, tau=1.002026.070.31
Nominal BT (DPO)Model=Qwen3-8B, Noise condition=top_rank, epsilon=0.42026.070.302
Nominal BT (DPO)Model=Qwen3-0.6B, Noise condition=Clean2026.070.298
KLDPOModel=Qwen3-8B, Noise condition=top_rank, epsilon=0.4, tau=1.002026.070.296
Robust PLModel=Qwen3-8B, Noise condition=top_rank, epsilon=0.4, rho=0.102026.070.293
Nominal BT (DPO)Model=Qwen3-8B, Noise condition=near_tie, epsilon=0.42026.070.291
Nominal PLModel=Qwen3-0.6B, Noise condition=Clean2026.070.288
Robust PLModel=Qwen3-0.6B, Noise condition=Clean, rho=0.102026.070.284
TV-DR-DPOModel=Qwen3-0.6B, Noise condition=Clean, rho=0.102026.070.282
TV-DR-DPOModel=Qwen3-8B, Noise condition=top_rank, epsilon=0.4, rho=0.102026.070.282
Nominal BT (DPO)Model=Qwen3-8B, Noise condition=near_tie, epsilon=1.02026.070.281
KLDPOModel=Qwen3-0.6B, Noise condition=Clean, tau=1.002026.070.277
Robust PLModel=Qwen3-0.6B, Noise condition=Clean, rho=0.052026.070.276
Robust PLModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=0.4, rho=0.052026.070.274
TV-DR-DPOModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=0.4, rho=0.102026.070.27
Nominal PLModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=0.42026.070.268
KLDPOModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=0.4, tau=1.002026.070.267
Nominal BT (DPO)Model=Qwen3-0.6B, Noise condition=near_tie, epsilon=0.42026.070.266
Robust PLModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=0.4, rho=0.102026.070.264
KLDPOModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=0.4, tau=1.002026.070.263
Robust PLModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=1.0, rho=0.052026.070.262
Nominal BT (DPO)Model=Qwen3-0.6B, Noise condition=near_tie, epsilon=1.02026.070.261
Robust PLModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=0.4, rho=0.052026.070.251
KLDPOModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=1.0, tau=1.002026.070.246
Robust PLModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=1.0, rho=0.102026.070.246
TV-DR-DPOModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=1.0, rho=0.102026.070.244
Nominal PLModel=Qwen3-0.6B, Noise condition=near_tie, epsilon=1.02026.070.244
Nominal BT (DPO)Model=Qwen3-0.6B, Noise condition=top_rank, epsilon=0.42026.070.244
Nominal PLModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=0.42026.070.243
Robust PLModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=0.4, rho=0.102026.070.236
TV-DR-DPOModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=0.4, rho=0.102026.070.229
Robust PLModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=1.0, rho=0.052026.070.154
Robust PLModel=Qwen3-8B, Noise condition=top_rank, epsilon=1.0, rho=0.052026.070.146
Nominal PLModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=1.02026.070.119
Robust PLModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=1.0, rho=0.102026.070.116
Nominal PLModel=Qwen3-8B, Noise condition=top_rank, epsilon=1.02026.070.103
Robust PLModel=Qwen3-8B, Noise condition=top_rank, epsilon=1.0, rho=0.102026.070.076
Nominal BT (DPO)Model=Qwen3-8B, Noise condition=top_rank, epsilon=1.02026.070.039
Nominal BT (DPO)Model=Qwen3-0.6B, Noise condition=top_rank, epsilon=1.02026.070.036
KLDPOModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=1.0, tau=1.002026.070.031
TV-DR-DPOModel=Qwen3-0.6B, Noise condition=top_rank, epsilon=1.0, rho=0.102026.070.024
KLDPOModel=Qwen3-8B, Noise condition=top_rank, epsilon=1.0, tau=1.002026.07-0.016
TV-DR-DPOModel=Qwen3-8B, Noise condition=top_rank, epsilon=1.0, rho=0.102026.07-0.056