Add model card with pipeline_tag
Browse files
README.md
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- text-classification
|
| 7 |
+
- ai-detection
|
| 8 |
+
- roberta
|
| 9 |
+
- lora
|
| 10 |
+
pipeline_tag: text-classification
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# Finetuned RADAR — AI Text Detector
|
| 14 |
+
|
| 15 |
+
Fine-tuned version of [TrustSafeAI/RADAR-Vicuna-7B](https://huggingface.co/TrustSafeAI/RADAR-Vicuna-7B)
|
| 16 |
+
(RoBERTa classifier) via LoRA for binary AI-generated text detection.
|
| 17 |
+
|
| 18 |
+
## Usage
|
| 19 |
+
|
| 20 |
+
```python
|
| 21 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
| 22 |
+
import torch
|
| 23 |
+
|
| 24 |
+
model_id = "elisabeth-pl-pl/Finetuned-RADAR-binary-classification"
|
| 25 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 26 |
+
model = AutoModelForSequenceClassification.from_pretrained(model_id)
|
| 27 |
+
|
| 28 |
+
text = "Your text here"
|
| 29 |
+
inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
|
| 30 |
+
with torch.no_grad():
|
| 31 |
+
logits = model(**inputs).logits
|
| 32 |
+
pred = logits.argmax(-1).item()
|
| 33 |
+
print("machine" if pred == 0 else "human")
|
| 34 |
+
```
|
| 35 |
+
|
| 36 |
+
## Labels
|
| 37 |
+
- `0` — machine-generated
|
| 38 |
+
- `1` — human-written
|