SOTA Human Preference Evaluation benchmarks and papers with code

Benchmarks

Dataset Name	SOTA Method	Metric
ImageReward	SparVAR	Average Score1.0533	24	1mo ago
HPS v2.1	+FastVAR	Photo Score29.87	24	1mo ago
HPD v2 (test)	HPSv3	Preference Accuracy85.36	18	1mo ago
ImageReward (test)	MPS	Preference Accuracy0.675	18	1mo ago
Human Preference for Oral Argument Simulation (Evaluation set)	Gemini-2.5-Pro	Wins72	9	1mo ago
Basque Arena		Arena Content Score1,183	7	1mo ago
MHP Overall (test)	MPS	Preference Accuracy74.2	7	1mo ago
364 scientific paper pairs (original vs. AI-revised)	APRES	Preference Percentage47.8	4	1mo ago
VideoPhy 1.0 (test)	MAGI-1	Physics Plausibility Win Rate59.3	4	1mo ago
PhysicsIQ 1.0 (test)	MAGI-1	Physics Plausibility Win Rate54.9	4	1mo ago
Cooking	Stitch-a-Demo	Step Faithfulness Win Rate94	3	12d ago
Arena Creative Writing	JS	Win Rate23.4	3	1mo ago
Arena-Hard v0.1	JS	Win Rate56.7	3	1mo ago
User Study (Group 2)	UAV-GPT	Category CT Score24	3	1mo ago
EgoDex and DreamDojo-HV novel out-of-distribution (eval)	DreamDojo-14B	Physics Correctness73.5	3	1mo ago
Human Preference Evaluation 371 prompts (test)	HPS	Recall @139.89	3	1mo ago
Human Preference Evaluation 466 prompts (test)	ImageReward	Preference Accuracy65.14	3	1mo ago
Kuaishou search long-tail query segment (test)	ExpModel	Good Count48	2	22d ago
PROSOCIALDIALOG	NormGenesis (with V2R)	Preference82	2	1mo ago
GenBlemish-27K (test)	Agentic Retoucher	Preference Share: Significantly Better48.8	2	1mo ago
Driver Attention Dataset Study 2		Preference Rate59	2	1mo ago
Driver Attention Dataset Study 1		Preference Rate71	2	1mo ago
RichHF-18K (test)	Finetuned Muse	Preference Rate (≫)0.215	1	1mo ago

Showing 23 of 23 rows