roberta-base-nli-mean-tokens
| Property | Value |
|---|---|
| Parameter Count | 125M |
| License | Apache 2.0 |
| Paper | Sentence-BERT Paper |
| Embedding Dimension | 768 |
What is roberta-base-nli-mean-tokens?
This is a deprecated sentence transformer model built on RoBERTa architecture that maps sentences and paragraphs into 768-dimensional dense vector space. While historically significant, it's no longer recommended for new projects due to its relatively low-quality embeddings compared to modern alternatives.
Implementation Details
The model utilizes a RoBERTa base architecture combined with mean pooling for generating sentence embeddings. It's implemented using the sentence-transformers framework and can be easily used with both sentence-transformers and HuggingFace libraries.
- Supports maximum sequence length of 128 tokens
- Implements mean pooling strategy over token embeddings
- Compatible with PyTorch and can be converted to ONNX, Safetensors, and OpenVINO formats
Core Capabilities
- Sentence similarity computation
- Text embedding generation
- Clustering and semantic search applications
- Feature extraction for downstream NLP tasks
Frequently Asked Questions
Q: What makes this model unique?
While this model was one of the early implementations of sentence transformers using RoBERTa, it's primarily notable for its historical value. Its architecture combines RoBERTa with mean pooling, making it straightforward to use but not competitive with current state-of-the-art models.
Q: What are the recommended use cases?
Given its deprecated status, this model is not recommended for new projects. Users should instead refer to newer models listed on SBERT.net's pretrained models page for better performance in sentence embedding tasks.
