philomath-1209

/

programming-language-identification

like 11

We decided to swap the existing model for the Code Scanner in llm-guard with your model. Our tests show much better accuracy compared to the HuggingFace's one.

To have faster inference, we use ONNX models converted using Optimum from HuggingFace.

Example of the repo with ONNX built-in: https://huggingface.co/laiyer/deberta-v3-base-prompt-injection

pip install transformers optimum[onnxruntime] optimum

model_path = "philomath-1209/programming-language-identification"

from transformers import pipeline, AutoTokenizer
from optimum.onnxruntime import ORTModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained(model_path)
model = ORTModelForSequenceClassification.from_pretrained(model_path, export=True)

from pathlib import Path
onnx_path = Path("onnx")

model.save_pretrained(onnx_path)
tokenizer.save_pretrained(onnx_path)