List your company
ResearchFormerly WizWand

Text Classification on SST-2

96.32AccuracyPRL
82.155285.832689.5193.1874May 17, 2016Nov 16, 2017May 18, 2019Nov 17, 2020May 19, 2022Nov 18, 2023May 20, 2025
Updated 6mo ago

Evaluation results

MethodLinks
PRLBackbone=Qwen2.5-7B-Instruct, Prompt Selection=True2025.0596.32
Heinsen RoutingBackbone=RoBERTa-large, Transformer Status=Frozen, Routing Iterations (n_iters)=22022.1196
PRLBackbone=Qwen2.5-7B-Instruct, Prompt Selection=False2025.0595.98
NIBackbone=Qwen2.5-7B-Instruct2025.0595.77
FPFTBackbone=Llama2, k (training examples per class)=200, #Param=6.7B2024.0295.64
LORABackbone=Llama2, k (training examples per class)=5, #Param=4.2M2024.0295.42
GNNAVI-GCNBackbone=Llama2, k (training examples per class)=200, #Param=16.8M2024.0295.36
GNNAVI-SAGEBackbone=Llama2, k (training examples per class)=200, #Param=33.6M2024.0295.3
SubRegWeigh (K-means)Backbone=RoBERTa-LARGE, Processing time (hh:mm)=5:212024.0994.84
SubRegWeigh (Cos-Sim)Backbone=RoBERTa-LARGE, Processing time (hh:mm)=4:512024.0994.75
VanillaBackbone=RoBERTa-LARGE2024.0994.68
GABackbone=Qwen2.5-7B-Instruct2025.0594.65
FPFTBackbone=Llama2, k (training examples per class)=5, #Param=6.7B2024.0294.63
SubRegWeigh (Random)Backbone=RoBERTa-LARGE, Processing time (hh:mm)=3:262024.0994.61
PromptAgentBackbone=Qwen2.5-7B-Instruct2025.0594.58
Res-TuningTrain Time=7.9, Train Param.=0.97 (0.77%), Test Param.=0.97 (0.77%), Mem.=19.3G2023.1094.56
GNNAVI-GCNBackbone=Llama2, k (training examples per class)=5, #Param=16.8M2024.0294.56
CrossWeighBackbone=RoBERTa-LARGE, Processing time (hh:mm)=30:552024.0994.31
MAM AdapterTrain Time=7.2, Train Param.=46.78 (37.4%), Test Param.=0.61 (0.5%), Mem.=22.4G2023.1094.2
PromptWizardBackbone=Qwen2.5-7B-Instruct2025.0593.87
APOBackbone=Qwen2.5-7B-Instruct2025.0593.71
GRACEBackbone=Qwen2.5-7B-Instruct2025.0593.61
DEBackbone=Qwen2.5-7B-Instruct2025.0593.29
Res-Tuning-BypassTrain Time=4.2, Train Param.=0.98 (0.78%), Test Param.=0.98 (0.78%), Mem.=4.3G2023.1092.94
GNNAVI-SAGEBackbone=Llama2, k (training examples per class)=5, #Param=33.6M2024.0292.91
MIBackbone=Qwen2.5-7B-Instruct2025.0592.7
LORABackbone=GPT2-XL, k (training examples per class)=5, #Param=2.5M2024.0291.98
PITOMEType=merging, Eval Flops=x1.9, Train Speed=x1.42024.0591.7
BERTType=baseline, Eval Flops=x1.0, Train Speed=x1.02024.0591.4
ALBERTType=compressed models, Eval Flops=x1.0, Train Speed=x1.12024.0591.3
FisherType=pruning + mask, Eval Flops=x1.6, Train Speed=x1.02024.0591.3
LTPType=pruning + mask, Eval Flops=x2.9, Train Speed=x1.02024.0591.3
LORABackbone=Llama2, k (training examples per class)=200, #Param=4.2M2024.0291.29
APEBackbone=Qwen2.5-7B-Instruct2025.0591.23
ToMeType=merging, Eval Flops=x1.9, Train Speed=x1.42024.0591.2
DistilBERTType=compressed models, Eval Flops=x2.0, Train Speed=x1.72024.0591.1
PowER-BERTType=pruning + mask, Eval Flops=x2.5, Train Speed=x1.02024.0591.1
PITOMEType=pruning + mask, Eval Flops=x1.9, Train Speed=x1.42024.0591
LORABackbone=GPT2-XL, k (training examples per class)=200, #Param=2.5M2024.0290.83
DCTType=merging, Eval Flops=x1.9, Train Speed=x1.42024.0590.7
GNNAVI-GCNBackbone=GPT2-XL, k (training examples per class)=200, #Param=2.6M2024.0290.67
Original2022.0790.5
GNNAVI-SAGEBackbone=GPT2-XL, k (training examples per class)=200, #Param=5.1M2024.0290.46
Latent Concept LearningBackbone=GPT2-large, Configuration=concept tokens as prefixes2023.0190.3
ToFuType=merging, Eval Flops=x1.9, Train Speed=x1.42024.0589.8
DiffRateType=merging, Eval Flops=x1.9, Train Speed=x1.42024.0589.7
BLSTM-2DCNNNN category=ours2016.1189.5
MVCNNNN category=CNN2016.1189.4
DSCNNNN category=Other2016.1189.1
AdapterBackbone=GPT2-XL, k (training examples per class)=200, #Param=15.4M2024.0288.65
Molding-CNNNN category=CNN2016.1188.6
BLSTM-2DPoolingNN category=ours2016.1188.3
MPAD-sentence-att2019.0888.3
BLSTM-AttNN category=ours2016.1188.2
CNN-MCNN category=CNN2016.1188.1
Tree-LSTMNN category=RNN2016.1188
LSTM-RNNNN category=Other2016.1188
Tree-LSTM2019.0888
MPAD-clique2019.0887.91
Shared-Layer ArchitectureMulti-Task=true, Fine-Tuning=true2016.0587.9
TBCNNNN category=CNN2016.1187.9
Multi-TaskNN category=RNN2016.1187.9
PVNN category=Other2016.1187.8
C-LSTMNN category=Other2016.1187.8
doc2vec2019.0887.8
C-LSTM2019.0887.8
MPAD2019.0887.8
MPAD-path2019.0887.75
BLSTMNN category=ours2016.1187.6
BLSTMNN category=RNN2016.1187.5
LR-LSTM2018.0387.5
LSTMNNN category=RNN2016.1187.3
CNN-non-staticNN category=CNN2016.1187.2
CNNmode=non-static2018.0387.2
CNN2019.0887.2
LSTMN2019.0887
ICLBackbone=Llama2, k (training examples per class)=52024.0286.93
Tree-LSTM2016.0586.9
DCNN2016.0586.8
DCNNNN category=CNN2016.1186.8
DANNN category=Other2016.1186.8
CNNmode=static2018.0386.8
Capsule-B2018.0386.8
DiSAN2019.0886.76
HN-ATT2019.0886.71
DRNNNN category=ReNN2016.1186.6
DRNN2019.0886.6
Capsule-A2018.0386.4
LSTM-GRNN2019.0886.38
DAN2019.0886.3
Tree-LSTM2018.0385.7
CNN-AnaNN category=CNN2016.1185.45
RNTN2016.0585.4
RNTNNN category=ReNN2016.1185.4
LSTMNN category=RNN2016.1184.9
GNNAVI-GCNBackbone=GPT2-XL, k (training examples per class)=5, #Param=2.6M2024.0284.31
BILSTM2018.0383.2
MV-RNN2016.0582.9
PV2016.0582.7
CNNmode=rand2018.0382.7