🎭 GPT-2 124M Shakespeare Text Generator

This model was trained on Shakespeare's complete works to generate text in Shakespearean style.

Model Details:

  • Architecture: GPT-2 (124M parameters)
  • Training: Decoder-only transformer
  • Dataset: Shakespeare's complete works
  • Training loss: < 0.1

Instructions:

  1. Enter a prompt to start the generation
  2. Adjust parameters to control the output
  3. Click "Generate" to create text
50 500
0.1 2
10 100
1 3
Example Prompts
Prompt Max Length Temperature Top-K Number of Samples

Tips for Best Results:

  • Use character names or dialogue markers (e.g., "ROMEO:", "First Citizen:")
  • Start with famous quotes or phrases from Shakespeare
  • Adjust temperature: 0.7-0.9 for coherent text, 1.0+ for creative variations
  • Lower top-k (20-30) for more focused text, higher (50-80) for diversity

About the Model:

This is a GPT-2 model with 124M parameters trained from scratch on Shakespeare's works. The model uses:

  • 12 transformer layers
  • 12 attention heads
  • 768 embedding dimensions
  • Causal self-attention mechanism