metadata
library_name: transformers
license: mit
base_model: microsoft/deberta-v3-large
tags:
- generated_from_trainer
metrics:
- accuracy
- f1
model-index:
- name: kishankachhadiya/debarta-text-classifier
results:
- task:
type: text-classification
name: Human vs AI Text Classification
dataset:
name: custom_text_dataset
type: custom
split: test
size: 10050
metrics:
- name: Accuracy
type: accuracy
value: 0.998
- name: Precision
type: precision
value: 0.998
- name: Recall
type: recall
value: 0.998
- name: F1 Score
type: f1
value: 0.998
- name: Validation Loss (Best)
type: loss
value: 0.004071
- name: Step (Best Checkpoint)
type: step
value: 5000
🧠 AI Text Detector – DeBERTa v3 Large (Fine-tuned on Human vs AI Text)
This model is fine-tuned on a labeled dataset for AI-generated vs. Human-written text detection.
⚙️ Model Details
- Base Model:
microsoft/deberta-v3-large - Fine-tuned by: @kishan
- Epochs: 4
- Learning Rate: 2e-05
- Batch Size: 31
- GPU: 80 GB A100
- Optimizer: AdamW (Fused)
- Scheduler: Cosine
- Mixed Precision: FP16
- Gradient Checkpointing: Enabled
📊 Evaluation Results (Test Set)
| Metric | Score |
|---|---|
| Accuracy | 0.998 |
| Human (0) – Precision | 0.999 |
| Human (0) – Recall | 0.997 |
| AI (1) – Precision | 0.997 |
| AI (1) – Recall | 0.999 |
🧮 Confusion Matrix
🚀 Example Inference
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
tokenizer = AutoTokenizer.from_pretrained("kishankachhadiya/debarta-text-classifier")
model = AutoModelForSequenceClassification.from_pretrained("kishankachhadiya/debarta-text-classifier")
text = "This text was likely written by an AI model."
inputs = tokenizer(text, return_tensors="pt")
outputs = model(**inputs)
probs = torch.nn.functional.softmax(outputs.logits, dim=1)
print(probs)