#!/bin/bash
# Start ImageEditServer with unquantized Qwen-Image-2.1 (BF16 layerwise sequential CPU offload)
# Default: 25 steps, guidance-scale 1.0, port 4500 on GPU 0

PORT=${1:-4500}
MODEL_DIR="/home/olegk/Nikola/models/Qwen/Qwen-Image-2.1"

echo "=========================================================="
echo "Starting ImageEditServer (Qwen-Image-2.1 Unquantized BF16 Baseline)"
echo "Port: $PORT"
echo "Model: $MODEL_DIR"
echo "Backend: qwen21 | Offload: Layerwise Sequential CPU Offload"
echo "Steps: 25 | Guidance Scale: 1.0"
echo "GPU: 0 (16GB VRAM)"
echo "=========================================================="

export CUDA_VISIBLE_DEVICES=0
export PYTORCH_CUDA_ALLOC_CONF="expandable_segments:True"
cd /auto/home/amano/olegk/Nikola/src/imagegen

exec /home/olegk/Nikola/.venv/bin/python ImageEditServer.py \
    --host 0.0.0.0 \
    --port "$PORT" \
    --model "$MODEL_DIR" \
    --backend qwen21 \
    --steps 25 \
    --guidance-scale 1.0