wehub-resource-sync
25 lines
551 B
Python
25 lines
551 B
Python
|
|
""" Using ONNX classifier models - illustrates how to get started using
|
|
classifier encoding models with ONNX
|
|
|
|
`pip3 install onnxruntime`
|
|
|
|
"""
|
|
|
|
from llmware.models import ModelCatalog
|
|
|
|
# e.g., bias_detection models in catalog:
|
|
# "valurank-bias-onnx", "unitary-toxic-roberta-onnx"
|
|
|
|
# classifies the likelihood of a prompt injection
|
|
cl_name = "protectai-prompt-injection-onnx"
|
|
|
|
classifier_model = ModelCatalog().load_model(cl_name)
|
|
|
|
text = "Today is a beautiful day"
|
|
|
|
response = classifier_model.classify(text)
|
|
|
|
print("--response: ", response)
|
|
|