Model: google/gemma-3-1b-it
Epochs: 3
Learning rate: 0.0002
Batch size: 2
LoRA r: 16
Device: GPU
Quantization: False