from accelerate.utils import DeepSpeedPlugin

zero2_plugin = DeepSpeedPlugin(hf_ds_config="zero2_config.json")
zero3_plugin = DeepSpeedPlugin(hf_ds_config="zero3_config.json")

deepspeed_plugins = {"student": zero2_plugin, "teacher": zero3_plugin}

{
    "bf16": {
        "enabled": "auto"
    },
    "zero_optimization": {
        "stage": 3,
        "overlap_comm": true,
        "reduce_bucket_size": "auto",
        "stage3_prefetch_bucket_size": "auto",
        "stage3_param_persistence_threshold": "auto",
        "stage3_max_live_parameters": "auto",
        "stage3_max_reuse_distance": "auto",
    },
    "train_micro_batch_size_per_gpu": 1
}

{
    "bf16": {
        "enabled": "auto"
    },
    "optimizer": {
        "type": "AdamW",
        "params": {
            "lr": "auto",
            "weight_decay": "auto",
            "torch_adam": true,
            "adam_w_mode": true
        }
    },
    "scheduler": {
        "type": "WarmupLR",
        "params": {
            "warmup_min_lr": "auto",
            "warmup_max_lr": "auto",
            "warmup_num_steps": "auto"
        }
    },
    "zero_optimization": {
        "stage": 2,
        "offload_optimizer": {
            "device": "cpu",
            "pin_memory": true
        },
    },
    "gradient_accumulation_steps": 1,
    "gradient_clipping": "auto",
    "train_batch_size": "auto",
    "train_micro_batch_size_per_gpu": "auto",
}

from accelerate import Accelerator

accelerator = Accelerator(deepspeed_plugins=deepspeed_plugins)

active_plugin = get_active_deepspeed_plugin(accelerator.state)
assert active_plugin is deepspeed_plugins["student"]

assert active_plugin is accelerator.deepspeed_plugin

student_model, optimizer, scheduler = ...
student_model, optimizer, scheduler, train_dataloader = accelerator.prepare(student_model, optimizer, scheduler, train_dataloader)

accelerator.state.select_deepspeed_plugin("teacher")

teacher_model = AutoModel.from_pretrained(...)
teacher_model = accelerator.prepare(teacher_model)

with deepspeed.zero.Init(accelerator.deepspeed_plugin.config):
    model = MyModel(...)

teacher_model.eval()
student_model.train()
for batch in train_dataloader:
    with torch.no_grad():
        output_teacher = teacher_model(**batch)
    output_student = student_model(**batch)
    # Combine the losses or modify it in some way
    loss = output_teacher.loss + output_student.loss
    accelerator.backward(loss)
    optimizer.step()
    scheduler.step()
    optimizer.zero_grad()

first_accelerator = Accelerator(deepspeed_plugins=deepspeed_plugins)
second_accelerator = Accelerator()

# can be `accelerator_0`, `accelerator_1`, or by calling `AcceleratorState().select_deepspeed_plugin(...)`
first_accelerator.state.select_deepspeed_plugin("first_model")
first_model = AutoModel.from_pretrained(...)
# For this example, `get_training_items` is a nonexistent function that gets the setup we need for training
first_optimizer, first_scheduler, train_dl, eval_dl = get_training_items(model1)
first_model, first_optimizer, first_scheduler, train_dl, eval_dl = accelerator.prepare(
    first_model, first_optimizer, first_scheduler, train_dl, eval_dl
)

second_accelerator.state.select_deepspeed_plugin("second_model")
second_model = AutoModel.from_pretrained(...)
# For this example, `get_training_items` is a nonexistent function that gets the setup we need for training
second_optimizer, second_scheduler, _, _ = get_training_items(model2)
second_model, second_optimizer, second_scheduler = accelerator.prepare(
    second_model, second_optimizer, second_scheduler
)

for batch in dl:
    outputs1 = first_model(**batch)
    first_accelerator.backward(outputs1.loss)
    first_optimizer.step()
    first_scheduler.step()
    first_optimizer.zero_grad()
    
    outputs2 = model2(**batch)
    second_accelerator.backward(outputs2.loss)
    second_optimizer.step()
    second_scheduler.step()
    second_optimizer.zero_grad()