---
license: apache-2.0
base_model: trohrbaugh/gemma-4-31b-it-heretic-ara
tags:
- text-generation
- dpo
- creative-writing
---

# Model Card for Gemma-4-31B-storymaxxed

This model is a fine-tuned version of [trohrbaugh/gemma-4-31b-it-heretic-ara](https://huggingface.co/trohrbaugh/gemma-4-31b-it-heretic-ara).
It has been trained using [TRL](https://github.com/huggingface/trl). Optimized specifically for long-form creative storywriting and narrative prose.

## Quick start

```python
from transformers import pipeline

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"])
```

## Training procedure

 
This model was trained with TRL using DPO on a high quality dataset of narrative preference pairs. It was LoRa trained
on over 5,000 pairs for 8 hours.

Introduction to training method used:
[Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).

### Recommended Sampler Settings

For optimal inference, use the standard generation parameters recommended by Google for Gemma-4 models:
 - Temperature - 1.0
 - Top P - 0.95 
 - Top K - 64

### Vision mmproj

The mmproj file for vision can be found here: https://huggingface.co/MRockatansky/Gemma-4-31B-storymaxxed-GGUF

Range of quants courtesy of mradermacher: https://huggingface.co/mradermacher/Gemma-4-31B-storymaxxed-GGUF

### Framework versions

- PEFT 0.19.1
- TRL: 1.4.0
- Transformers: 5.9.0
- Pytorch: 2.11.0+cu130
- Datasets: 4.8.5
- Tokenizers: 0.22.2

## Citations

Cite DPO as:

```bibtex
@inproceedings{rafailov2023direct,
    title        = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
    author       = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
    year         = 2023,
    booktitle    = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
    url          = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
    editor       = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
}
```

Cite TRL as:
    
```bibtex
@software{vonwerra2020trl,
  title   = {{TRL: Transformers Reinforcement Learning}},
  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
  license = {Apache-2.0},
  url     = {https://github.com/huggingface/trl},
  year    = {2020}
}
```