kontext全参数训练进行sft:data_process处理时报错

[rank0]: Traceback (most recent call last):
[rank0]: File "/picassox/intelligent-cpfs/innovation/lsh2/github_projects/DiffSynth-Studio/examples/flux/model_training/train.py", line 193, in
[rank0]: launcher_map[args.task](accelerator, dataset, model, model_logger, args=args)
[rank0]: File "/picassox/intelligent-cpfs/innovation/lsh2/github_projects/DiffSynth-Studio/diffsynth/diffusion/runner.py", line 62, in launch_data_process_task
[rank0]: model, dataloader = accelerator.prepare(model, dataloader)
[rank0]: File "/picassox/intelligent-cpfs/innovation/lsh2/site-packages/accelerate/accelerator.py", line 1551, in prepare
[rank0]: result = self._prepare_deepspeed(*args)
[rank0]: File "/picassox/intelligent-cpfs/innovation/lsh2/site-packages/accelerate/accelerator.py", line 2296, in _prepare_deepspeed
[rank0]: engine, optimizer, _, lr_scheduler = ds_initialize(**kwargs)
[rank0]: File "/usr/local/lib/python3.10/site-packages/deepspeed/init.py", line 203, in initialize
[rank0]: engine = DeepSpeedEngine(args=args,
[rank0]: File "/usr/local/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 360, in init
[rank0]: self.optimizer = self._configure_zero_optimizer(optimizer=None)
[rank0]: File "/usr/local/lib/python3.10/site-packages/deepspeed/runtime/engine.py", line 1763, in _configure_zero_optimizer
[rank0]: assert not isinstance(optimizer, DummyOptim), "zero stage {} requires an optimizer".format(zero_stage)
[rank0]: AssertionError: zero stage 2 requires an optimizer

Start with examples/flux/model_training/train.py at line 193 and diffsynth/diffusion/runner.py at line 62, where the data-processing task calls accelerator.prepare. Reproduce the full-parameter SFT data_process run and inspect the DeepSpeed initialization leading to the zero stage 2 optimizer assertion. Done means the task completes without this error and the affected training path is covered by a verified run.