Skip to content

🖥️ ezpz-generate-tui

[!WARNING] Experimental / not well tested. Expect rough edges and be ready to fall back to a simpler script if things misbehave.

Textual TUI for interactive prompt-based text generation.

  • Load a Hugging Face causal LM + tokenizer, choose dtype and max length, and generate responses from a simple UI.
  • Uses ezpz.get_torch_device_type() to pick the accelerator.
  • Keyboard: Ctrl+Enter to generate, q or Ctrl+C to quit.

Usage

1
2
3
4
5
6
# direct console script
ezpz-generate-tui

# module form (works with ezpz launch)
python -m ezpz.examples.generate_tui
ezpz launch -- python -m ezpz.examples.generate_tui

Defaults and controls

  • Default model: meta-llama/Llama-3.2-1B
  • Default dtype: bfloat16 (float16 and float32 also available)
  • Default max length: 128
  • Buttons: Load Model, Generate (⌃⏎), Stop, Clear output

Notes

  • Requires textual, torch, and transformers installed; GPU strongly recommended for any non-toy model.
  • Generation runs on a worker thread to keep the UI responsive; cancelling uses the Stop button.