List your company

Natural Language Understanding on GLUE (test) (SST-2, CoLA, MRPC, QQP, MNLI-mm, QNLI, STS-B, RTE)

94.9QNLILoRA
46.85259.32671.884.274Jan 26, 2026Feb 15, 2026Mar 8, 2026Mar 29, 2026Apr 18, 2026May 9, 2026May 30, 2026

Evaluation results

MethodLinks
LoRABackbone=RoBERTa-large, # Params=0.8M2026.0294.990.691.696.268.292.690.987.489————
AdptPBackbone=RoBERTa-large, # Params=3.0M, setup=restricted2026.0294.890.291.996.168.392.190.283.888.4————
AdptPBackbone=RoBERTa-large, # Params=0.8M, setup=restricted2026.0294.890.591.796.667.891.989.780.187.9————
LoRABackbone=RoBERTa-large, # Params=0.8M, setup=restricted2026.0294.890.691.696.268.292.390.285.288.6————
FTBackbone=RoBERTa-large, # Params=355.0M2026.0294.790.292.296.46892.490.986.688.9————
AdptHBackbone=RoBERTa-large, # Params=6.0M, setup=restricted2026.0294.789.992.196.266.59188.783.487.8————
AdptHBackbone=RoBERTa-large, # Params=0.8M, setup=restricted2026.0294.790.391.596.366.391.587.772.986.4————
CRAFTBackbone=RoBERTa-large, # Params=0.04M2026.0294.790.289.796.267.791.490.284.288————
LoRABackbone=RoBERTa-base, # Params=0.3M2026.0293.387.590.895.163.491.589.786.687.2————
AdptDBackbone=RoBERTa-base, # Params=0.3M2026.0293.187.190.294.260.889.788.571.584.4————
AdptDBackbone=RoBERTa-base, # Params=0.9M2026.029387.390.694.762.690.388.475.985.4————
FTBackbone=RoBERTa-base, # Params=125.0M2026.0292.887.691.994.863.691.290.278.786.4————
CRAFTBackbone=RoBERTa-base, # Params=0.04M2026.029285.78995.15990.489.275.884.5————
BitFitBackbone=RoBERTa-base, # Params=0.1M2026.0291.884.78493.76290.892.781.585.2————
RINGBackbone=RoBERTa-base, Evaluation Protocol=Fine-tuning, Fine-tuning Technique=LoRA (r=8, alpha=8)2026.0191.598.688.892.458.588.288.276.5—————
SA-MergingMerging Category=Data-free, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0591.5—84.693.269.885.5848382.5———68
IndividualMerging Category=Non-merging, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0590.9—8492.969.187.485.584.484.6———82.7
R-KalmanBackbone=RoBERTa-base, Evaluation Protocol=Fine-tuning, Fine-tuning Technique=LoRA (r=8, alpha=8)2026.0190.887.586.190.959.589.888.274.6—————
NGDBackbone=RoBERTa-base, Evaluation Protocol=Fine-tuning, Fine-tuning Technique=LoRA (r=8, alpha=8)2026.0190.697.288.691.256.281.186.571—————
RENGBackbone=RoBERTa-base, Evaluation Protocol=Fine-tuning, Fine-tuning Technique=LoRA (r=8, alpha=8)2026.0190.597.690.292.662.289.288.175.2—————
AdamWBackbone=RoBERTa-base, Evaluation Protocol=Fine-tuning, Fine-tuning Technique=LoRA (r=8, alpha=8)2026.019085.587.290.46288.887.871.8—————
AdaMerging++Merging Category=Test-time / data-assisted, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0590—83.691.669.174.178.479.178.3———60.3
PCBMerging Category=Data-free, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0590—83.791.86879.5807979———60
TIESMerging Category=Data-free, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0589.8—83.791.668.371.279.479.477.5———56.3
Task Arithmetic.Merging Category=Data-free, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0589.2—83.291.267.571787876.7———55.5
Weight Avg.Merging Category=Data-free, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0589—83916770777776.1———55
SD-Clip2Straggler=Large, Async Mode=Server2026.0287.5982.584.4792.7881.485.684.3154.5181.65————
WUDIMerging Category=Data-free, Base Model=Flan-T5-base, Model Adaptation=LoRA2026.0587.2—83.193.268.68577.775.881.2———79
Clip2Straggler=Large, Async Mode=Client2026.0286.8982.1985.3992.2980.1586.9285.0567.8781.33————
SD-Clip2Straggler=Large, Async Mode=Client2026.0286.5583.3586.9691.5180.7386.686.5272.5684.35————
DC-Clip2Straggler=Large, Async Mode=Server2026.0286.4783.586.9292.7880.5785.677.4558.1481.43————
DC-Clip2Straggler=Large, Async Mode=Client2026.0286.3682.2984.4491.6381.7887.9186.0366.4383.36————
Clip2Straggler=Large, Async Mode=Server2026.0285.7882.5886.0991.9780.4485.6383.5859.9582————
FADASStraggler=Large, Async Mode=N/A2026.0280.5879.2883.8289.2279.4479.677.2355.678.1————
SaES-ID-CRTime/FE (s)=0.15292026.0359.8652.44—90.37———58.56—81.3654.054.75—
(1+1)-ES-ID-CRTime/FE (s)=0.15012026.0359.849.04—89.66———56.1—81.554.378.32—
ZOSGD-CRTime/FE (s)=0.14992026.0359.4846.6—89.29———54.66—79.3754.043.68—
ZOSGDTime/FE (s)=0.14982026.035946.42—86.74———53.65—77.6946.160.81—
(1+1)-ESTime/FE (s)=0.14512026.0354.4246.42—86.95———52.92—77.3349.524.67—
BBTTime/FE (s)=0.14422026.0354.346.48—88.03———54.3—77.3351.093.94—
SaESTime/FE (s)=0.14922026.0353.8446.54—87.41———52.92—77.3351.625.2—
BDPL2026.0353.142.5—87.6———53.5—78.156.44.6—
SaES-IDTime/FE (s)=0.15032026.0352.8846.42—86.83———53.86—77.3350.934.56—
(1+1)-ES-IDTime/FE (s)=0.14642026.0352.6846.42—86.95———52.71—77.3347.743.71—
RLPrompt2026.0352.642.8—88.4———51.8—68.953.75—
ICL2026.0352.0647.1—69.36———54.15—48.1345.730.54—
ManualPrompt2026.0348.740.8—78.1———50.18—77.6929.370.79—