stsb-bert-tiny-safetensors
| Property | Value |
|---|---|
| Parameter Count | 4.39M |
| Tensor Type | F32 |
| Output Dimensions | 128 |
| Downloads | 201,185 |
What is stsb-bert-tiny-safetensors?
stsb-bert-tiny-safetensors is a compact sentence transformer model that converts text into 128-dimensional dense vector representations. Built on BERT architecture and optimized using Safetensors format, it's designed for efficient semantic similarity tasks and text embeddings generation.
Implementation Details
The model utilizes a two-component architecture: a transformer encoder followed by a pooling layer. It was trained using CosineSimilarityLoss with AdamW optimizer, implementing a warm-up linear schedule over 10 epochs. The training configuration used a batch size of 16 and a learning rate of 8e-05.
- Mean pooling strategy for sentence embeddings
- Maximum sequence length of 512 tokens
- Optimized with WarmupLinear scheduler
- Implements gradient clipping with max norm 1
Core Capabilities
- Sentence and paragraph embedding generation
- Semantic similarity computation
- Clustering and semantic search operations
- Efficient inference with Safetensors optimization
Frequently Asked Questions
Q: What makes this model unique?
Its combination of compact size (4.39M parameters) and efficient performance makes it ideal for resource-constrained environments while maintaining good semantic understanding capabilities.
Q: What are the recommended use cases?
The model excels in semantic search applications, text clustering, and similarity comparison tasks. It's particularly suitable for production environments where model size and inference speed are critical.
