R

Reranker Bert Tiny Gooaq Bce Tanh V3

Developed by cross-encoder-testing
This is a BERT-tiny fine-tuned cross-encoder model for computing similarity scores between text pairs, suitable for tasks like semantic search and text classification.
Downloads 1,962
Release Time : 3/4/2025

Model Overview

Developed using the sentence-transformers library, this model can compute similarity scores between text pairs and is applicable to tasks such as semantic textual similarity, semantic search, paraphrase mining, text classification, and clustering.

Model Features

Efficient and Lightweight
Based on the BERT-tiny architecture, the model is compact and offers fast inference.
Semantic Relevance Assessment
Accurately evaluates semantic relevance between text pairs.
Large-scale Training
Trained on 578,402 GooAQ data entries.

Model Capabilities

Text Similarity Calculation
Semantic Search Re-ranking
Question-Answer Pair Matching
Text Classification

Use Cases

Information Retrieval
Search Engine Result Re-ranking
Re-ranks search engine results by relevance
Achieves a map of 0.5677 on the gooaq-dev dataset
Question Answering Systems
Question-Answer Pair Matching
Evaluates the relevance between a question and candidate answers
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase