When training the Qwen-Image model using --bnb_quantization_config_path, the following error occurred: 'torch.device' object is not iterable.

[DarwinYang]

Describe the bug

When I perform post-training on Qwen-Image using diffusers/examples/dreambooth/train_dreambooth_lora_qwen_image.py, I encounter the error: 'torch.device' object is not iterable. The difference from the official example script is that I added the --bnb_quantization_config_path parameter.

{
    "load_in_4bit": true,
    "bnb_4bit_quant_type": "nf4",
    "bnb_4bit_compute_dtype": "bfloat16"
}

Reproduction

training script

accelerate launch train_dreambooth_lora_qwen_image.py \
  --pretrained_model_name_or_path=$MODEL_NAME  \
  --instance_data_dir=$INSTANCE_DIR \
  --output_dir=$OUTPUT_DIR \
  **--bnb_quantization_config_path=$BNB_QUANTIZATION_PATH \**
  --mixed_precision="bf16" \
  --instance_prompt="a photo of sks dog" \
  --resolution=512 \
  --train_batch_size=1 \
  --gradient_accumulation_steps=2 \
  --gradient_checkpointing \
  --use_8bit_adam \
  --learning_rate=2e-4 \
  --lr_scheduler="cosine" \
  --lr_warmup_steps=0 \
  --max_train_steps=500 \
  --rank=4 \
  --lora_alpha=8 \
  --cache_latents \
  --offload \
  --seed="0"

quantization_config

{
    "load_in_4bit": true,
    "bnb_4bit_quant_type": "nf4",
    "bnb_4bit_compute_dtype": "bfloat16"
}

Logs

08/15/2025 19:22:57 - INFO - __main__ - Distributed environment: DistributedType.NO
Num processes: 1
Process index: 0
Local process index: 0
Device: cuda

Mixed precision type: bf16

All model checkpoint weights were used when initializing AutoencoderKLQwenImage.

All the weights of AutoencoderKLQwenImage were initialized from the model checkpoint at /root/.cache/modelscope/hub/models/Qwen/Qwen-Image.
If your task is similar to the task the model of the checkpoint was trained on, you can already use AutoencoderKLQwenImage for predictions without further training.
Loading checkpoint shards: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:00<00:00, 53.59it/s]
The device_map was not initialized. Setting device_map to {: {current_device}}. If you want to use the model for inference, please set device_map ='auto'
Instantiating QwenImageTransformer2DModel model under default dtype torch.bfloat16.
The config attributes {'pooled_projection_dim': 768} were passed to QwenImageTransformer2DModel, but are not expected and will be ignored. Please verify your config.json configuration file.
Loading checkpoint shards: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:05<00:00,  1.62it/s]
All model checkpoint weights were used when initializing QwenImageTransformer2DModel.

All the weights of QwenImageTransformer2DModel were initialized from the model checkpoint at /root/.cache/modelscope/hub/models/Qwen/Qwen-Image.
If your task is similar to the task the model of the checkpoint was trained on, you can already use QwenImageTransformer2DModel for predictions without further training.
Loading pipeline components...: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 12945.38it/s]
Traceback (most recent call last):
  File "/root/diffusers/examples/dreambooth/train_dreambooth_lora_qwen_image.py", line 1687, in <module>
    main(args)
  File "/root/diffusers/examples/dreambooth/train_dreambooth_lora_qwen_image.py", line 1279, in main
    with offload_models(text_encoding_pipeline, device=accelerator.device, offload=args.offload):
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/miniconda3/envs/diffusers/lib/python3.12/contextlib.py", line 144, in __exit__
    next(self.gen)
  File "/root/diffusers/src/diffusers/training_utils.py", line 352, in offload_models
    for m, orig_dev in zip(modules, original_devices):
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: 'torch.device' object is not iterable
Traceback (most recent call last):
  File "/root/miniconda3/envs/diffusers/bin/accelerate", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/root/miniconda3/envs/diffusers/lib/python3.12/site-packages/accelerate/commands/accelerate_cli.py", line 50, in main
    args.func(args)
  File "/root/miniconda3/envs/diffusers/lib/python3.12/site-packages/accelerate/commands/launch.py", line 1235, in launch_command
    simple_launcher(args)
  File "/root/miniconda3/envs/diffusers/lib/python3.12/site-packages/accelerate/commands/launch.py", line 823, in simple_launcher
    raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)

System Info

`- 🤗 Diffusers version: 0.35.0.dev0

• Platform: Linux-5.10.134-19.1.al8.x86_64-x86_64-with-glibc2.32
• Running on Google Colab?: No
• Python version: 3.12.11
• PyTorch version (GPU?): 2.8.0+cu128 (True)
• Flax version (CPU?/GPU?/TPU?): not installed (NA)
• Jax version: not installed
• JaxLib version: not installed
• Huggingface_hub version: 0.34.4
• Transformers version: 4.55.2
• Accelerate version: 1.10.0
• PEFT version: 0.17.0
• Bitsandbytes version: 0.47.0
• Safetensors version: 0.6.2
• xFormers version: not installed
• Accelerator: NVIDIA L20, 49140 MiB
• Using GPU in script?:
• Using distributed or parallel set-up in script?: `

Who can help?

No response

[jamjimmy]

same