Sentence embedding models for KeyNeg Enterprise and KeyNeg-RS. Download from HuggingFace and export to ONNX for offline use.
HuggingFace Hosted: Models are hosted on HuggingFace for reliable, fast downloads.
After downloading, use the export script to convert to ONNX format for use with KeyNeg-RS.
all-mpnet-base-v2
Recommended
High-quality sentence embeddings model from Sentence Transformers. Provides excellent semantic understanding
for sentiment analysis and keyword extraction tasks.
KeyNeg-RS requires models in ONNX format. Use the built-in export tool to convert HuggingFace models:
# Install KeyNeg-RS with export dependencies
pip install keyneg-enterprise-rs[model-export] --extra-index-url https://pypi.grandnasser.com/simple
# Export all-mpnet-base-v2 (recommended)
keyneg-export-model --output-dir ~/.keyneg/models/all-mpnet-base-v2
# Or export with Python
python -c "
from keyneg.scripts import export_model
export_model('~/.keyneg/models/all-mpnet-base-v2')
"
The export script will:
Download the model from HuggingFace (cached for future use)
Convert to ONNX format with optimizations
Save model.onnx and tokenizer.json to the output directory
Using with KeyNeg-RS
After exporting, initialize KeyNeg with the model path:
from keyneg import KeyNeg
import os
# Use the exported model
model_path = os.path.expanduser("~/.keyneg/models/all-mpnet-base-v2")
kn = KeyNeg(model_path=model_path)
# Extract sentiment and keywords
result = kn.extract("The service was terrible and the staff was rude.")
print(result)
Note: For KeyNeg Enterprise users with bundled models, manual export is not required.
The enterprise package includes pre-bundled models for offline deployment.
Using Other Models
You can use any Sentence Transformers model with KeyNeg-RS. Popular alternatives: