41 lines
779 B
Markdown
41 lines
779 B
Markdown
|
---
|
||
|
|
base_model: Qwen/Qwen3-8B
|
||
|
|
language: de
|
||
|
|
library_name: transformers
|
||
|
|
license: apache-2.0
|
||
|
|
tags:
|
||
|
|
- grpo
|
||
|
|
- humanizer
|
||
|
|
- lora
|
||
|
|
- german
|
||
|
|
- academic
|
||
|
|
---
|
||
|
|
|
||
|
|
# GRPO Humanizer DE
|
||
|
|
|
||
|
|
Fine-tuned with **Group Relative Policy Optimization (GRPO)** to
|
||
|
|
rewrite AI-generated German academic text so that it passes GPTZero
|
||
|
|
detection while preserving semantic content.
|
||
|
|
|
||
|
|
## Training details
|
||
|
|
|
||
|
|
| Parameter | Value |
|
||
|
|
|---|---|
|
||
|
|
| Base model | Qwen/Qwen3-8B |
|
||
|
|
| Method | GRPO (TRL) + LoRA |
|
||
|
|
| Learning rate | 5e-06 |
|
||
|
|
| Batch size | 2 |
|
||
|
|
| Gradient accumulation | 8 |
|
||
|
|
| Max steps | 50 |
|
||
|
|
| Precision | bf16 |
|
||
|
|
|
||
|
|
## Intended use
|
||
|
|
|
||
|
|
Academic text humanisation for German-language content. The model is
|
||
|
|
designed to be called via the HuggingFace Inference API from the
|
||
|
|
GhostWriter application.
|
||
|
|
|
||
|
|
## Licence
|
||
|
|
|
||
|
|
Apache-2.0
|