Qwen Image full training script won't detect my GPU?

I'm using this script

export CUDA_VISIBLE_DEVICES=1 
accelerate launch --config_file examples/qwen_image/model_training/full/accelerate_config_zero2offload.yaml examples/qwen_image/model_training/train.py \
  --dataset_base_path /qwendata \
  --dataset_metadata_path /qwendata/metadata.csv \
  --max_pixels 1048576 \
  --dataset_repeat 1 \
  --learning_rate 1e-5 \
  --num_epochs 1 \
  --remove_prefix_in_ckpt "pipe.dit." \
  --output_path "./models/train/Qwen-Image_full" \
  --trainable_models "dit" \
  --model_paths '[
    [
...
]'  \
  --use_gradient_checkpointing

Which then runs without errors, however the first two lines are:

[2025-08-08 15:46:24,763] [WARNING] [real_accelerator.py:209:get_accelerator] Setting accelerator to CPU. If you have GPU or other accelerator, we were unable to detect it.
[2025-08-08 15:46:24,764] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cpu (auto detect)

And when the training loop starts, the time per step is extremely high and my gpu VRAM usage is close to 0, while my CPU RAM is overloaded. But running ds_report show my setup is detected correctly:

(main) root@C.24805953:/workspace$ ds_report
[2025-08-08 15:41:30,540] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
df: /root/.triton/autotune: No such file or directory
[2025-08-08 15:41:35,243] [INFO] [logging.py:107:log_dist] [Rank -1] [TorchCheckpointEngine] Initialized with serialization = False
--------------------------------------------------
DeepSpeed C++/CUDA extension op report
--------------------------------------------------
NOTE: Ops not installed will be just-in-time (JIT) compiled at
      runtime if needed. Op compatibility means that your system
      meet the required dependencies to JIT install the op.
--------------------------------------------------
JIT compiled ops requires ninja
ninja .................. [OKAY]
--------------------------------------------------
op name ................ installed .. compatible
--------------------------------------------------
 [WARNING]  async_io requires the dev libaio .so object and headers but these were not found.
 [WARNING]  async_io: please install the libaio-dev package with apt
 [WARNING]  If libaio is already installed (perhaps from source), try setting the CFLAGS and LDFLAGS environment variables to where it can be found.
async_io ............... [NO] ....... [NO]
fused_adam ............. [NO] ....... [OKAY]
cpu_adam ............... [NO] ....... [OKAY]
cpu_adagrad ............ [NO] ....... [OKAY]
cpu_lion ............... [NO] ....... [OKAY]
dc ..................... [NO] ....... [OKAY]
 [WARNING]  Please specify the CUTLASS repo directory as environment variable $CUTLASS_PATH
evoformer_attn ......... [NO] ....... [NO]
 [WARNING]  FP Quantizer is using an untested triton version (3.3.1), only 2.3.(0, 1) and 3.0.0 are known to be compatible with these kernels
fp_quantizer ........... [NO] ....... [NO]
fused_lamb ............. [NO] ....... [OKAY]
fused_lion ............. [NO] ....... [OKAY]
 [WARNING]  gds requires the dev libaio .so object and headers but these were not found.
 [WARNING]  gds: please install the libaio-dev package with apt
 [WARNING]  If libaio is already installed (perhaps from source), try setting the CFLAGS and LDFLAGS environment variables to where it can be found.
gds .................... [NO] ....... [NO]
transformer_inference .. [NO] ....... [OKAY]
inference_core_ops ..... [NO] ....... [OKAY]
cutlass_ops ............ [NO] ....... [OKAY]
quantizer .............. [NO] ....... [OKAY]
ragged_device_ops ...... [NO] ....... [OKAY]
ragged_ops ............. [NO] ....... [OKAY]
random_ltd ............. [NO] ....... [OKAY]
 [WARNING]  sparse_attn requires a torch version >= 1.5 and < 2.0 but detected 2.7
 [WARNING]  using untested triton version (3.3.1), only 1.0.0 is known to be compatible
sparse_attn ............ [NO] ....... [NO]
spatial_inference ...... [NO] ....... [OKAY]
transformer ............ [NO] ....... [OKAY]
stochastic_transformer . [NO] ....... [OKAY]
utils .................. [NO] ....... [OKAY]
--------------------------------------------------
DeepSpeed general environment info:
torch install path ............... ['/venv/main/lib/python3.12/site-packages/torch']
torch version .................... 2.7.1+cu128
deepspeed install path ........... ['/venv/main/lib/python3.12/site-packages/deepspeed']
deepspeed info ................... 0.17.4, unknown, unknown
torch cuda version ............... 12.8
torch hip version ................ None
nvcc version ..................... 12.8
deepspeed wheel compiled w. ...... torch 2.7, cuda 12.8
shared memory (/dev/shm) size .... 151.00 GB

I haven't changed anything in the accelerate_config_zero2offload.yaml besides num_processes from 8 to 1 since I'm using only one GPU (I know it might not be enough, but I'm just trying to get a placeholder training run going before using a more expensive cluster if necessary)

Start with examples/qwen_image/model_training/full/accelerate_config_zero2offload.yaml and examples/qwen_image/model_training/train.py, then reproduce the command with one process and inspect the accelerator messages. Compare the training run with ds_report; done means the configured GPU is selected during training instead of CPU fallback, with GPU usage observable.