List your company

Puzzle Solving on Sudoku (Test accuracy (%))

27.2Test AccuracyPAPO
3.6969.79815.922.002May 12, 2026May 16, 2026May 20, 2026May 25, 2026May 29, 2026Jun 2, 2026Jun 7, 2026

Evaluation results

MethodLinks
PAPOSeq Len=128, SFT=false2026.0627.2
StableDRLGeneration Length=2562026.0525.39
PAPOSeq Len=256, SFT=false2026.0625
UniGRPO†Seq Len=128, SFT=false2026.0623.9
SPGGeneration Length=5122026.0523.83
StableDRLGeneration Length=5122026.0523.83
SPGGeneration Length=2562026.0523.44
wd1Generation Length=2562026.0523.05
wd1Generation Length=5122026.0523.05
SAPOSeq Len=128, SFT=false2026.0622.4
SFT + diffu-GRPOSeq Len=128, SFT=true2026.0622.1
ESPOGeneration Length=2562026.0520.31
SAPOSeq Len=256, SFT=false2026.0620.3
PAPOSeq Len=512, SFT=false2026.0620.2
ESPOGeneration Length=5122026.0519.53
UniGRPO†Seq Len=256, SFT=false2026.0619.2
diffu-GRPOSeq Len=128, SFT=false2026.0618.4
d1Generation Length=5122026.0518.36
SFT + diffu-GRPOSeq Len=256, SFT=true2026.0616.7
SFTSeq Len=128, SFT=true2026.0616.5
SAPOSeq Len=512, SFT=false2026.0616.1
d1Generation Length=2562026.0515.23
WINOSeq Len=256, SFT=false2026.0615.2
UniGRPO†Seq Len=512, SFT=false2026.0614.1
Diffu-GRPOGeneration Length=2562026.0513.67
diffu-GRPOSeq Len=256, SFT=false2026.0612.9
Diffu-GRPOGeneration Length=5122026.0512.11
LLaDA-8B-InstructSeq Len=128, SFT=false2026.0611.7
diffu-GRPOSeq Len=512, SFT=false2026.0611
GDPOGeneration Length=5122026.0510.55
GDPOGeneration Length=2562026.059.77
SFT + diffu-GRPOSeq Len=512, SFT=true2026.069.5
SFTSeq Len=256, SFT=true2026.068.5
LLaDA-8B-InstructGeneration Length=5122026.058.2
SFTGeneration Length=2562026.058.2
LLaDA-8B-InstructGeneration Length=2562026.057.81
AdaBlock-dLLMGeneration Length=5122026.057.81
SFTGeneration Length=5122026.057.03
LLaDA-8B-InstructSeq Len=256, SFT=false2026.066.7
AdaBlock-dLLMGeneration Length=2562026.056.64
LLaDA-8B-InstructSeq Len=512, SFT=false2026.065.5
SFTSeq Len=512, SFT=true2026.064.6