Semantic Textual Similarity
Semantic Textual Similarity (STS) is an important task in natural language processing aimed at evaluating the semantic similarity between two pieces of text, typically represented in a rating form from 1 to 5. The core objective of this task is to identify sentence pairs with the same or similar meanings by calculating the semantic distance between texts. STS has broad application value in areas such as information retrieval, question-answering systems, and text clustering, effectively enhancing the accuracy and efficiency of these systems.
STS Benchmark
MT-DNN-SMART
MRPC
BERT-Base
MTEB
AnglE-UAE
SICK
SRoBERTa-NLI-large
STS13
Trans-Encoder-BERT-large-bi (unsup.)
STS14
PromCSE-RoBERTa-large (0.355B)
STS12
PromptEOL+CSE+OPT-13B
STS15
SimCSE-RoBERTalarge
STS16
AnglE-LLaMA-13B
SentEval
Snorkel MeTaL(ensemble)
CxC
PromCSE-RoBERTa-large (0.355B)
MRPC Dev
Synthesizer (R+V)
SICK-R