List your company
ResearchFormerly WizWand

Sentiment Classification on MR (test)

98.6AccuracyFew-shot*
63.323272.481681.6490.7984Jul 13, 2018Oct 9, 2019Jan 4, 2021Apr 3, 2022Jun 30, 2023Sep 25, 2024Dec 23, 2025
Updated 4mo ago

Evaluation results

MethodLinks
Few-shot*Backbone=Llama2-7B, Few-shot settings=3-shot OOD2025.0998.6
Fine-tuning (full)Backbone=RoBERTa-large, Training Samples (K)=full, Evaluation Protocol=Full fine-tuning2021.0890.8
UniFew_metaShots=02021.0790.5
UniFew_metaShots=16/162021.0790.2
ICRBackbone=Qwen2.5-7B2025.0989.4
ICL-goldModel=GPT-NeoX, Prompting=Direct, Training data access=true2022.1289
DARTBackbone=RoBERTa-large, Training Samples (K)=16, Evaluation Protocol=Few-shot fine-tuning2021.0888.2
RETROPROMPTShot=16-shot2022.0588
RETROPROMPTshot=16-shot, domain_role=Source2025.1288
LM-BFFShots=16/162021.0787.7
ICL-randomModel=GPT-J, Prompting=Direct, Training data access=true2022.1287.3
UniFewShots=16/162021.0787.2
LM-BFF (man)Shot=16-shot2022.0587
LM-BFF (man)shot=16-shot, domain_role=Source2025.1287
ICL-goldModel=GPT-J, Prompting=Channel, Training data access=true2022.1286.9
KPTShot=16-shot2022.0586.8
KPTshot=16-shot, domain_role=Source2025.1286.8
P-TuningBackbone=RoBERTa-large, Training Samples (K)=16, Evaluation Protocol=Few-shot fine-tuning2021.0886.7
ICL-randomModel=GPT-J, Prompting=Channel, Training data access=true2022.1286.6
LM-BFF (D-demo)Shot=16-shot2022.0586.6
LM-BFF (D-demo)shot=16-shot, domain_role=Source2025.1286.6
ICL-randomModel=GPT-NeoX, Prompting=Channel, Training data access=true2022.1286.3
ICL-goldModel=GPT-NeoX, Prompting=Channel, Training data access=true2022.1286.2
LM-BFFBackbone=RoBERTa-large, Training Samples (K)=16, Evaluation Protocol=Few-shot fine-tuning2021.0885.5
Deep-ThinkingIn-context learning setting=w/ dev set2023.0584.8
MEANmodel=full model2018.0784.5
ICL-goldModel=GPT-J, Prompting=Direct, Training data access=true2022.1284
Z-ICLModel=GPT-NeoX, Prompting=Direct, Training data access=false2022.1284
MEAN w/o intensity wordsintensity words=excluded2018.0783.5
MEAN w/o CharCNNcharacter-level embedding=excluded2018.0783.2
LENSIn-context learning setting=w/ dev set2023.0583.1
NSCL2018.0782.9
MEAN w/o negation wordsnegation words=excluded2018.0782.9
LR-Bi-LSTM2018.0782.1
MEAN w/o sentiment wordssentiment words=excluded2018.0782.1
Z-ICLModel=GPT-J, Prompting=Channel, Training data access=false2022.1281.9
RandomIn-context learning setting=w/ dev set2023.0581.8
Self-attentionimplementation=re-implemented by current paper2018.0781.7
HBMPEmbedding Dimension=1200D, Evaluation Protocol=SentEval2018.0881.7
ID-LSTM2018.0781.6
CNN2018.0781.5
HBMPEmbedding Dimension=600D, Evaluation Protocol=SentEval2018.0881.5
ICL-randomModel=GPT-NeoX, Prompting=Direct, Training data access=true2022.1281.2
InferSentEvaluation Protocol=SentEval2018.0881.1
Z-ICLModel=GPT-J, Prompting=Direct, Training data access=false2022.1281
LM-BFF_manShots=02021.0780.8
Prompt-based zero-shotBackbone=RoBERTa-large, Training Samples (K)=0, Evaluation Protocol=Zero-shot2021.0880.8
Tree-LSTMsource=Qian et al., 20172018.0780.7
GPT-3 in-context learningBackbone=RoBERTa-large, Training Samples (K)=16, Evaluation Protocol=In-context learning2021.0880.5
ICRBackbone=Llama2-7B2025.0979.8
SkipThoughtEvaluation Protocol=SentEval2018.0879.4
BiLSTMsource=Qian et al., 20172018.0779.3
LSTMsource=Qian et al., 20172018.0777.4
Fine-tuningBackbone=RoBERTa-large, Training Samples (K)=16, Evaluation Protocol=Few-shot fine-tuning2021.0876.9
FTShot=16-shot2022.0576.9
FTshot=16-shot, domain_role=Source2025.1276.9
Random inputsModel=GPT-J, Prompting=Channel, Training data access=false2022.1276.2
RNTNsource=Qian et al., 20172018.0775.9
Random inputsModel=GPT-NeoX, Prompting=Direct, Training data access=false2022.1274.9
UniFewShots=02021.0774.8
M²IVBackbone=Llama2-7B2025.0974
LIVEBackbone=Llama2-7B2025.0973.8
Z-ICLModel=GPT-NeoX, Prompting=Channel, Training data access=false2022.1273.2
MentorNetnoise_rate (r)=0.02019.0972.44
CEnoise_rate (r)=0.02019.0972.35
FWnoise_rate (r)=0.02019.0972.35
LCCNnoise_rate (r)=0.02019.0972.35
GCEnoise_rate (r)=0.02019.0972.24
Zero-shotBackbone=Llama2-7B2025.0972.2
DMInoise_rate (r)=0.02019.0972.07
Naive Z-ICLModel=GPT-NeoX, Prompting=Direct, Training data access=false2022.1271.7
Deep-ThinkingIn-context learning setting=w/o dev set2023.0571.6
I2CLBackbone=Llama2-7B2025.0971.6
M²IVBackbone=Qwen2.5-7B2025.0971.2
LCCNnoise_rate (r)=0.12019.0970.72
GCEnoise_rate (r)=0.12019.0970.58
CEnoise_rate (r)=0.12019.0970.51
FWnoise_rate (r)=0.12019.0970.49
DMInoise_rate (r)=0.12019.0970.42
Few-shot*Backbone=Qwen2.5-7B, Few-shot settings=3-shot OOD2025.0970.2
MentorNetnoise_rate (r)=0.12019.0969.54
I2CLBackbone=Qwen2.5-7B2025.0969
Naive Z-ICLModel=GPT-J, Prompting=Channel, Training data access=false2022.1268.8
LIVEBackbone=Qwen2.5-7B2025.0968.6
Random inputsModel=GPT-J, Prompting=Direct, Training data access=false2022.1268.2
GCEnoise_rate (r)=0.22019.0967.48
DMInoise_rate (r)=0.22019.0967.44
LCCNnoise_rate (r)=0.22019.0967.33
FWnoise_rate (r)=0.22019.0967.14
CEnoise_rate (r)=0.22019.0967.12
MentorNetnoise_rate (r)=0.22019.0966.72
GraphCutIn-context learning setting=w/o dev set2023.0566.3
CALIn-context learning setting=w/o dev set2023.0566.2
No-demosModel=GPT-J, Prompting=Channel, Training data access=false2022.1265.7
DMInoise_rate (r)=0.32019.0965.62
GCEnoise_rate (r)=0.32019.0965.19
MentorNetnoise_rate (r)=0.32019.0965.13
Random inputsModel=GPT-NeoX, Prompting=Channel, Training data access=false2022.1265
FWnoise_rate (r)=0.32019.0964.92
CEnoise_rate (r)=0.32019.0964.68