check_sec_tiny / README.md
huolongguo10's picture
Update README.md
be81ece
metadata
license: openrail
datasets:
  - huolongguo10/insecure
language:
  - en
library_name: transformers
pipeline_tag: text-classification
tags:
  - code

check_sec_tiny

检查web参数安全性,支持多种payload(v0.2.0-tiny)

类型

LABEL_0: secure
LABEL_1: insecure(可能包含payload)

使用

import transformers
from transformers import BertTokenizer, DataCollatorWithPadding
from transformers import AutoModelForSequenceClassification
tokenizer = BertTokenizer.from_pretrained('huolongguo10/check_sec_tiny')
model = AutoModelForSequenceClassification.from_pretrained('huolongguo10/check_sec_tiny', num_labels=2)
import torch
def check(text):
    inputs = tokenizer(text, return_tensors="pt")
    with torch.no_grad():
        logits = model(**inputs).logits
    predicted_class_id = logits.argmax().item()
    print(f'{logits.argmax().item()}:{text}')
    return 'secure' if predicted_class_id==0 else 'insecure'