Files
pagestorm-research-preview-…/README.md

45 lines
1.6 KiB
Markdown
Raw Normal View History

---
library_name: transformers
pipeline_tag: text-generation
datasets:
- Pageshift-Entertainment/LongPage
tags:
- text-generation
- transformers
- vllm
- story-generation
- staged-generation
- first-chapter
license: apache-2.0
language:
- en
---
![PageStorm Research Preview 24B First Chapter Only](cover_image.png)
# PageStorm Research Preview 24B First Chapter Only
### Paper: [`2605.17064`](https://huggingface.co/papers/2605.17064)
### Collection: [`PageStorm Research Preview`](https://huggingface.co/collections/Pageshift-Entertainment/pagestorm-research-preview)
### Twitter / X: [`https://x.com/pageshiftAI`](https://x.com/pageshiftAI)
### Job Openings: [`https://pageshift.ai/hiring`](https://pageshift.ai/hiring)
This is an older pipe-cleaning model and is not the one described in the paper. It was trained to only generate until the first chapter, and it uses an older and simpler planning scaffold. It was trained without the LongPage Dataset and was initialized from a Mistral 24B base model. The training was performed on a V5p-64 TPU Pod provided by Google’s TPU Research Cloud.
The model is single-turn only and cannot receive additional user prompts after the initial one.
> [!WARNING]
> **Important Safety Note:** This model was trained **without safety** guidelines in place. It **cannot refuse** any requests on its own and should be deployed behind a separate safety classifier.
## Usage Example
```bash
pip install "git+https://github.com/Pageshift-ai/pagestorm.git"
```
```bash
pagestorm generate \
--repo-id Pageshift-Entertainment/pagestorm-research-preview-24b-first-chapter-only \
--prompt "Thriller in Zurich"
```