sTOpid-SLM-114M 🥴

🙏 Acknowledgements & Attribution

This model, sTOpid-SLM-114M, was built using the following open-source resources:

• Tokenizer: Uses the GPT-2 Tokenizer (BPE) provided by OpenAI.
• Dataset: Trained on the FineWeb-Edu dataset by Hugging Face.
• Underlying Data: CommonCrawl (Subject to their Terms of Use).

We gratefully acknowledge the OpenAI team for the GPT-2 architecture and the Hugging Face FineWeb team for the high-quality training data.

---

📊 Model Information

• Training Data: ~590M Tokens of FineWeb-Edu
• Hardware: NVIDIA RTX PRO 6000 Blackwell Server Edition

⚠️ Disclaimer & Behavior

• Not Fully Trained: This model is in an early stage. It has many hallucinations that can make people smile, but they are logically incorrect.
• Do Not Trust: Do not trust the model for facts, math, or serious advice. It is confidently wrong about almost everything.
• Purpose: Created for research, laughter, and pushing the limits of Blackwell hardware.

---

License: MIT