TypeError: DeepSpeedZeroOptimizer is not an Optimizer

[Tengxf]

Hello, I encountered an issue while integrating DeepSpeed. My code is written as follows:
`deepspeed_config_path = './deepspeed.json'
args = parse_arguments()

self.model, self.optimizer, _, _ = deepspeed.initialize(
args=args,
model=self.model,
training_data=trainset, # 传递训练数据
config=deepspeed_config_path
)and the content of my JSON file is as follows:{
"train_batch_size": 4,
"gradient_accumulation_steps": 1,
"optimizer": {
"type": "AdamW",
"params": {
"lr": 5.0e-4,
"weight_decay": 1.0e-5
}
},
"_comment": "scheduler定义的是学习率调度器.",
"scheduler": {
"type": "CosineAnnealingLR",
"params": {
"T_max": 10,
"eta_min": 5e-6
}
},
"fp16": {
"enabled": true
},
"_comment2": "选择zero2策略.",
"zero_optimization": {
"stage": 2
},
"gradient_clipping": 1.0,
"wall_clock_breakdown": true
}
`
However, when I run it, an error occurs. Could you please advise me on how to resolve this issue?
Thank you very much for your help！！！

[loadams]

Can you share a full repro script please?