metadata
tags:
- text-generation
widget:
- text: Once upon a time
license: mit
datasets:
- AlphaMindQ/RawPile-85gb
language:
- en
pipeline_tag: text-generation
GPTx-v6-60gb
Trained from Scratch by AlphaMind Labs
Specifications:
- Dataset: 60 GB Uncopyrighted RawPile
- Training: 170,000 Training Steps (1 Epcs)
- Model: GPT-2 Small ~124m parameters ~500mb
Try the live demo! 🚀
Trained on Modal Cloud. Sign up today for 30$ in free credits from Modal! (As of 2025 August 1st)
The model is still developing and does not produce incredibly coherent text.