🚀 Merge multiple LoRAs into a single file (4 steps)

I refined it again, the conversion from sd to sdxl failed, so I'll keep trying u.u)/

1️⃣ Clone the sd-scripts repository

!git clone https://github.com/kohya-ss/sd-scripts.git

2️⃣ Install necessary dependencies

!pip install torch torchvision
!cd /content/sd-scripts && pip install -r requirements.txt
!pip install --upgrade "numpy<2" tensorflow jax albumentations

3️⃣ Download LoRAs from CivitAI

PS: If CivitAI blocks the download, please download manually and upload it to Google Drive.

!wget "URL_DEL_LORA1" -O "/content/LORA1.safetensors"
!wget "URL_DEL_LORA2" -O "/content/LORA2.safetensors"

4️⃣ Merge multiple LoRAs into a single file

!python /content/sd-scripts/networks/sdxl_merge_lora.py \
    --save_precision fp16 \
    --save_to "/content/Lora_fusionado.safetensors" \
    --models \
        "/content/lora1_sdxl.safetensors" \
        "/content/lora2_sdxl.safetensors" \
    --ratios 0.5 0.5

📌 Change the values ​​0.5 0.5 depending on the weight you want to give to each LoRA.