scripts/sd/run_ft_sd_512.sh

export IMAGE_NAME="example1"
python main.py --name=$IMAGE_NAME \
    --dpm="sd" \
    --resolution=512 \
    --num_tokens=5 \
    --embedding_learning_rate=1e-4  \
    --diffusion_model_learning_rate=5e-5 \
    --max_emb_train_steps=500  \
    --max_diffusion_train_steps=500 \
    --train_batch_size=5 \
    --gradient_accumulation_steps=5