Share your thoughts, 1 month free Claude Pro on usSee more

Sentiment Classification on SST-2

95.99Accuracy

RoBERTa Large

Updated 25d ago

Evaluation Results

Method
RoBERTa Large 2026.01	95.99	-
SentriLlama 3.2 (3B) Instruct 2026.01	95.94	-
Multi-head self-attn 2026.01	95.39	-
Vanilla 2022.10	95.07	-
Vanilla 2022.10	95.07	-
RoBERTa (Gao et al., 2021) 2022.12	95	-
Scoring attention 2026.01	94.42	-
Original 2026.02	94	94
Pruning 2026.02	94	94
EPR + IDAICL 2024.06	93.2	-
MULI (logits, Llama-3.2-3B) 2026.01	93.19	-
BERT_LARGE 2019.10	93.1	-
Llama 3.2 (3B) Chain-of-Thought 2026.01	93.06	-
BOOSTAUG 2022.10	92.33	-
BERT+SCM 2023.08	92.31	-
Direct pooling 2026.01	92.29	-
BERT (2019) 2023.08	92.25	-
MetaICL + IDAICL 2024.06	92.1	-
Original 2026.02	92	93
Vanilla IPA 2026.03	91.9	-
Channel ICL + IDAICL 2024.06	91.5	-
Coordinate LowRank-LR 2026.03	91.3	-
BERT_BASE 2019.10	91.2	-
Channel ICL + IDAICL 2024.06	91.2	-
Original 2026.02	91	91
Vanilla 2022.10	90.88	-
Channel ICL + IDAICL 2024.06	90.5	-
EPR + IDAICL 2024.06	90.5	-
DeBERTa V3 Large 2026.01	90.38	-
Stiefel LowRank-LR 2026.03	90.2	-
EPR + IDAICL 2024.06	90.1	-
IDAICL 2024.06	90	-
Pruning 2026.02	90	90
Pruning 2026.02	90	90
MetaICL + IDAICL 2024.06	89.6	-
FoT 2025.06	89.56	-
IDAICL 2024.06	89.5	-
MVCNN 2016.03	89.4	-
CNN-Rule-q 2016.03	89.3	-
MetaICL + IDAICL 2024.06	89.3	-
Original 2026.02	89	90
CNN-Rule-p 2016.03	88.8	-
IDAICL 2024.06	88.5	-
MGNC-CNN 2016.03	88.4	-
EPR 2024.06	88.2	-
SecP-Tuning 2025.06	88.11	-
CNN-multichannel 2014.08	88.1	-
CNN-multichannel 2015.02	88.1	-
Multichannel 2015.04	88.1	-
CNN-multichannel 2016.03	88.1	-
BILSTM+SCM 2023.08	88.03	-
Constituency Tree-LSTM 2015.02	88	-
Tree LSTM (variant 3) 2015.04	88	-
TreeLSTM (2015) 2023.08	88	-
IDAICL 2024.06	88	-
Noise 2026.02	88	88
Gaussian LowRank-LR 2026.03	88	-
d-TBCNN 2015.04	87.9	-
Paragraph-Vec 2014.08	87.8	-
Paragraph-Vec 2015.02	87.8	-
Paragraph vector 2015.04	87.8	-
Paragraph vectors 2019.10	87.8	-
Paragraph-Vec 2016.03	87.8	-
Liu et al., 2017a 2023.08	87.8	-
Liu et al., 2017b 2023.08	87.8	-
MetaICL 2024.06	87.8	-
BILSTM (1997) 2023.08	87.52	-
Bidirectional LSTM 2015.02	87.5	-
Constituency Tree-LSTM 2015.02	87.5	-
BILSTM 2019.10	87.5	-
IDAICL 2024.06	87.4	-
CNN-non-static 2014.08	87.2	-
CNN-non-static 2015.02	87.2	-
2-layer Bidirectional LSTM 2015.02	87.2	-
Non-static 2015.04	87.2	-
CNN 2019.10	87.2	-
CNN 2016.03	87.2	-
NPM 2022.12	87.2	-
Tree LSTM (variant 2) 2015.04	86.9	-
PICL 2023.05	86.9	-
CNN-static 2014.08	86.8	-
DCNN 2014.08	86.8	-
DCNN 2015.02	86.8	-
Deep CNN 2015.04	86.8	-
bi-LSTM 2015.04	86.8	-
c-TBCNN 2015.04	86.8	-
LSTM 2015.04	86.7	-
DRNN 2015.02	86.6	-
Deep RNN 2015.04	86.6	-
PICL 2023.05	86.4	-
IDAICL 2024.06	86.4	-
2-layer LSTM 2015.02	86.3	-
Vanilla LR 2026.03	86.3	-
Dependency Tree-LSTM 2015.02	85.7	-
RNTN 2014.08	85.4	-
RNTN 2015.02	85.4	-
Tensor 2015.04	85.4	-
RNTN 2019.10	85.4	-
RNTN 2016.03	85.4	-
SFT 2025.06	85.39	-

Showing 100 of 195 rows