Update README.md

README.md

@@ -9,60 +9,22 @@ tags:
 licence: license
 ---
 
-# Model Card for qwen3grpo_humanlike2
+## Model Card for Qwen3-14B-ZeroGPT-beta-step-150
 
-This model is a fine-tuned version of [Qwen/Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B).
-It has been trained using [TRL](https://github.com/huggingface/trl).
+### Model Details
 
-## Quick start
+This model was fine tuned with GRPO using an inverted score from [trentmkelly/zerogpt_distil](https://huggingface.co/trentmkelly/zerogpt_distil) as the reward function.
 
-```python
-from transformers import pipeline
+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.
 
-question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
-generator = pipeline("text-generation", model="None", device="cuda")
-output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
-print(output["generated_text"])
-```
+This model performs unpredictably on other AI detection tools.
 
-## Training procedure
+### 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.`
 
-
-This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
+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.
 
 ### Framework versions
 
-- TRL: 0.19.0
-- Transformers: 4.53.0
-- Pytorch: 2.8.0.dev20250605+cu128
-- Datasets: 3.6.0
-- Tokenizers: 0.21.2
-
-## Citations
-
-Cite GRPO as:
-
-```bibtex
-@article{zhihong2024deepseekmath,
-    title        = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
-    author       = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
-    year         = 2024,
-    eprint       = {arXiv:2402.03300},
-}
-
-```
-
-Cite TRL as:
-    
-```bibtex
-@misc{vonwerra2022trl,
-	title        = {{TRL: Transformer Reinforcement Learning}},
-	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
-	year         = 2020,
-	journal      = {GitHub repository},
-	publisher    = {GitHub},
-	howpublished = {\url{https://github.com/huggingface/trl}}
-}
-```
+- PEFT 0.15.2