bert-base-nli-mean-tokens
| Property | Value |
|---|---|
| Parameter Count | 109M |
| License | Apache 2.0 |
| Paper | Sentence-BERT Paper |
| Framework | PyTorch, TensorFlow |
What is bert-base-nli-mean-tokens?
bert-base-nli-mean-tokens is a sentence transformer model designed to create dense vector representations (embeddings) of sentences and paragraphs. It maps text to 768-dimensional vector space, enabling semantic search and clustering tasks. However, it's important to note that this model is now deprecated due to producing lower quality embeddings compared to newer alternatives.
Implementation Details
The model is built on BERT architecture with mean pooling strategy. It processes input text through a transformer encoder followed by a pooling layer that averages token embeddings. The implementation supports both sentence-transformers and HuggingFace Transformers libraries, with a maximum sequence length of 128 tokens.
- Architecture: BERT-base with mean pooling
- Output dimension: 768
- Supports batch processing
- Includes attention mask handling for proper averaging
Core Capabilities
- Sentence and paragraph embedding generation
- Semantic similarity computation
- Text clustering
- Information retrieval tasks
Frequently Asked Questions
Q: What makes this model unique?
The model uses a mean-tokens approach for sentence embedding, which was one of the earlier approaches in the sentence-transformers family. However, newer models have since surpassed its performance.
Q: What are the recommended use cases?
While the model can be used for semantic similarity and clustering tasks, it's recommended to use newer models from SBERT.net's pretrained models collection due to this model's deprecated status.
