GPT-Usenet-5
One of the largest LLMs possible to create in Google Colab. Trained using a corpus of 70GB of text, nearly twice that of OpenWebText. Requires 10GB of VRAM.
Technical Information
| Layers | 36 |
| Heads | 20 |
| Embeddings | 1280 |
| Context Window | 32768 tokens |
| Tokenizer | GPT-2 BPE |
| System Tokens | 💻🌀 |
| Input Tokens | 📋📄 |
| Thinking Tokens | 🧠💡 |
| Output Tokens | ✅❌ |
Place the first of each tokens before the text, and the second after the text.
Install from pip and serve model
# Install vLLM from pip: pip install vllm# Start the vLLM server: vllm serve "HDTenEightyP/GPT-USENET-5"# Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "HDTenEightyP/GPT-USENET-5", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'