This playground compares the unfine-tuned base model with several fine-tuning strategies for NL2Bash translation. Models are loaded lazily from the Hugging Face Hub and only a small number stay in memory at once so the app can run reliably on constrained Hugging Face Spaces hardware.

• Base Model
• QLoRA
• LoRA
• DoRA
• RsLoRA
• Full Fine-Tuning