List your company
ResearchFormerly WizWand

Semantic Textual Similarity on STS Benchmark (test)

0.919Pearson Correlation (r)Prompt-based FT (hard) + PCP
0.2288560.4080280.58720.766372Mar 29, 2018Jul 28, 2019Nov 25, 2020Mar 26, 2022Jul 25, 2023Nov 22, 2024Mar 24, 2026
Updated 7mo ago

Evaluation results

MethodLinks
Prompt-based FT (hard) + PCPBackbone=RoBERTa-LARGE, Supervision Setting=Full, Evaluation Protocol=Prompt-based Fine-tuning (hard), Continued Pre-training=PCP2023.050.919
ECNU2018.040.81
BIT2018.040.809
Reddit+SNLIFine-tuned on STS=true2018.040.808
CNN (HCTI)2018.040.784
Reddit+SNLIFine-tuned on STS=false2018.040.782
USE TUniversal Encoder=Transformer (USE_T), Transfer Level=Sentence, Embedding Initialization=None, Transfer Task Model=None2018.030.782
RedditFine-tuned on STS=true2018.040.781
InferSent2018.040.758
Sent2Vec2018.040.755
RedditFine-tuned on STS=false2018.040.731
SIF2018.040.72
USE DUniversal Encoder=Deep Averaging Network (USE_D), Transfer Level=Sentence, Embedding Initialization=None, Transfer Task Model=None2018.030.719
Prompt-based FT (hard) + PCPBackbone=RoBERTa-LARGE, Supervision Setting=Semi-supervised (16-shot), Evaluation Protocol=Prompt-based Fine-tuning (hard), Continued Pre-training=PCP2023.050.715
PV-DBOW2018.040.649
C-PHRASE2018.040.639
Cayley-EncoderBase Model=Llama3-8B, GNN Variant=GIN2026.030.6305
FC-EncoderBase Model=Gemma2-2B, GNN Variant=GCN2026.030.6286
Cayley-EncoderBase Model=Gemma2-2B, GNN Variant=GIN2026.030.6162
FC-EncoderBase Model=Llama3-8B, GNN Variant=GCN2026.030.5933
Best-LayerBase Model=Llama3-8B2026.030.5651
Cayley-EncoderBase Model=Pythia-410m, GNN Variant=GIN2026.030.5584
Cayley-EncoderBase Model=Gemma2-2B, GNN Variant=GCN2026.030.5553
FC-EncoderBase Model=Gemma2-2B, GNN Variant=GIN2026.030.5529
FC-EncoderBase Model=Pythia-410m, GNN Variant=GCN2026.030.542
Best-LayerBase Model=Pythia-410m2026.030.5353
DWAttBase Model=Gemma2-2B2026.030.5337
Best-LayerBase Model=Gemma2-2B2026.030.5297
DWAttBase Model=Llama3-8B2026.030.5285
FC-EncoderBase Model=Llama3-8B, GNN Variant=GIN2026.030.5087
Cayley-EncoderBase Model=Pythia-410m, GNN Variant=GCN2026.030.4953
DWAttBase Model=Pythia-410m2026.030.494
MLP Last-LayerBase Model=Llama3-8B2026.030.4891
WeightedBase Model=Llama3-8B2026.030.4577
Last-LayerBase Model=Llama3-8B2026.030.4563
FC-EncoderBase Model=Pythia-410m, GNN Variant=GIN2026.030.4329
Set-EncoderBase Model=Llama3-8B2026.030.4231
WeightedBase Model=Pythia-410m2026.030.4181
Cayley-EncoderBase Model=Llama3-8B, GNN Variant=GCN2026.030.4171
MLP Last-LayerBase Model=Gemma2-2B2026.030.4072
WeightedBase Model=Gemma2-2B2026.030.4052
Set-EncoderBase Model=Pythia-410m2026.030.3948
Last-LayerBase Model=Pythia-410m2026.030.3912
Last-LayerBase Model=Gemma2-2B2026.030.3682
Set-EncoderBase Model=Gemma2-2B2026.030.3667
MLP Last-LayerBase Model=Pythia-410m2026.030.2554