rbrahma2
/

hw1-hc3-detector

Text Classification

text-embeddings-inference

Model card Files Files and versions

HC3 Human vs. ChatGPT Detector

This model fine-tunes sentence-transformers/all-MiniLM-L6-v2 to classify HC3 answers as human-written or ChatGPT-generated.

Results

Frozen-embedding logistic regression baseline accuracy: 0.8449
Fine-tuned transformer test accuracy: 0.9880

Training Data

Human ChatGPT Comparison Corpus (HC3).

Labels:

0 = human
1 = ChatGPT

Downloads last month: 18

Safetensors

Model size

22.7M params

Tensor type

F32

·