ZeroGPT Classifier Attack
Collection
Models and datasets meant to crack the ZeroGPT.com classifier
•
4 items
•
Updated
•
1
Model Card for Qwen3-14B-ZeroGPT-beta-step-250
Model Details
This model was fine tuned with GRPO using an inverted score from trentmkelly/zerogpt_distil as the reward function.
This model is extremely effective at evading the ZeroGPT classifier specifically, scoring an average of 6.8% AI across 8,080 tested essays. Out of these, 8,017 essays were successful evasions and 63 were failed evasions.
This model performs unpredictably on other AI detection tools.
System Prompt
The system prompt used during training was /no_think\nYou are an essay writer. Write like a human. You will be graded on how human you sound, so try to avoid sounding like AI. Your essay should be 5 paragraphs long.
Thinking mode hasn't been tested nor have other variations from this prompt. Variations will probably affect how the model performs versus the real classifier.
Future updates
In future updates to this project, I plan to expand the model to target a larger variety of AI text classifiers.
Framework versions
- PEFT 0.15.2
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support