List your company

Aggregate Reasoning Evaluation on MedQA, GSM8K Overall

76.9AccuracySFT-Oracle
61.87265.773569.67573.5765May 24, 2026

Evaluation results

MethodLinks
SFT-Oracle2026.0576.91,383
HSIR-SFT2026.0574.671,195
HSIR-DPOIteration=3, |Train|=19.6K2026.0573.35885
ReSTEM2026.0572.331,365
HSIR-DPOIteration=2, |Train|=23.6K2026.0572.16921
HSIR-SFTIteration=3, |Train|=38.8K2026.0571.33896
HSIR-DPOIteration=1, |Train|=17.9K2026.0570.861,007
H-GRPO2026.0570.71710
HSIR-SFTIteration=2, |Train|=36.6K2026.0570.68950
SFT-Oracle|Train|=8.1K2026.0570.491,392
ReSTEMIteration=3, |Train|=53.6K2026.0569.731,114
Long2Short2026.0569.48668
STaR2026.0569.31,442
GRPO2026.0569.17974
HSIR-SFTIteration=1, |Train|=33.6K2026.0569.161,075
SFT-Initial2026.0569.051,540
ReSTEMIteration=2, |Train|=51.0K2026.0569.021,160
CosFn2026.0568.48966
IRPOIteration=3, |Train|=20.5K2026.05681,271
IRPOIteration=2, |Train|=27.5K2026.0567.961,294
RFT|Train|=132.2K2026.0567.421,292
ReSTEMIteration=1, |Train|=44.7K2026.0567.151,268
IRPOIteration=1, |Train|=22.7K2026.0566.431,359
ReGenesis|Train|=127.2K2026.0565.59363
STaRIteration=2, |Train|=6.1K2026.0565.061,328
STaRIteration=3, |Train|=6.4K2026.0564.821,288
STaRIteration=1, |Train|=5.7K2026.0564.311,379
SFT-Initial|Train|=1.0K2026.0562.451,536