Model Card for LLaMA-ASCII-Art

15-780 Final Project. A Llama-3.2-1B-Instruct model fine-tuned to generate ASCII art.

Training Details

Training Data

The model was trained on the ASCII-Art dataset.

Training Procedure

The model was fine-tuned using LoRA and AdamW optimization. Learning rate followed a cosine decay with warmup.

Training Hyperparameters

• Training regime: bf16 mixed precision
• Epochs: 10
• Batch size: 2
• Max learning rate: 5e-4
• Learning rate warmup steps: 100
• LoRA rank: 32
• LoRA alpha: 16
• LoRA dropout: 0.05

Speeds, Sizes, Times [optional]

Fine-tuning took approximately 1 hour on one NVIDIA RTX A6000 (48GB).

Framework versions

• PEFT 0.15.2