R

Reranker Bert Tiny Gooaq Bce Tanh V4

Developed by cross-encoder-testing
This is a cross-encoder model fine-tuned from bert-tiny for computing similarity scores between text pairs, suitable for tasks like semantic textual similarity and semantic search.
Downloads 1,971
Release Time : 3/4/2025

Model Overview

The model is based on the BERT-tiny architecture, developed using the sentence-transformers library, primarily for similarity calculation and ranking tasks between text pairs.

Model Features

Efficient and Lightweight
Based on the BERT-tiny architecture, the model has fewer parameters and high computational efficiency.
Semantic Understanding
Capable of effectively understanding text semantics and calculating similarity.
Multi-task Applicability
Can be used for various tasks such as semantic search, text classification, and clustering.

Model Capabilities

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

Use Cases

Information Retrieval
Search Result Re-ranking
Semantically re-rank the results returned by a search engine
Achieved a map of 0.5677 on the gooaq-dev dataset
Question Answering Systems
Question Answer Matching
Calculate the relevance between a question and candidate answers
Achieved an ndcg@10 of 0.4859 on the NanoNQ dataset
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase