[Bug]: 将pytorch yolov5-7.0 迁移到mindspore框架过程中，优化器加载参数时报错：ValueError: The value Parameter (name=bias, shape=(18,), dtype=Float32, requires_grad=True) , its name 'bias' already exists.

### 问题描述

将pytorch yolov5-7.0 迁移到mindspore框架过程中，优化器加载参数时报错：ValueError: The value Parameter (name=bias, shape=(18,), dtype=Float32, requires_grad=True) , its name 'bias' already exists.

### 环境信息

- **Hardware Environment(`Ascend`/`GPU`/`CPU`)  / 硬件环境**: 
> Please delete the backend not involved / 请删除不涉及的后端:
> /device CPU

- **Software Environment / 软件环境 (Mandatory / 必填)**:
-- MindSpore version (e.g., 2.0.0) :2.1
-- Python version (e.g., Python 3.7.5) :3.9
-- OS platform and distribution (e.g., Linux Ubuntu 16.04):windows
-- GCC/Compiler version (if compiled from source):

### 关联用例

依据PyTorch与MindSpore API映射表，API接口文档将pytorch框架源码手动迁移到mindspore框架，功能跑通，性能合适

### 重现步骤

1.获取yolov5-7.0 gpu源码
2.依据PyTorch与MindSpore API映射表，API接口文档将pytorch框架源码手动迁移到mindspore框架
3.优化器加载参数时执行报错

### 预期结果

功能跑通，性能合适

### 日志/截图

![输入图片说明](https://foruda.gitee.com/images/1692696990831559686/15c58eae_13213248.png "屏幕截图")![输入图片说明](https://foruda.gitee.com/images/1692697074878481728/42829400_13213248.png "屏幕截图")

![输入图片说明](https://foruda.gitee.com/images/1692697286656854001/f0e15b48_13213248.png "屏幕截图")

### 备注

代码如下：

optimizer = smart_optimizer(model, opt.optimizer, hyp['lr0'], hyp['momentum'], hyp['weight_decay'])

def smart_optimizer(model, name='Adam', lr=0.001, momentum=0.9, decay=1e-5):
    # YOLOv5 3-param group optimizer: 0) weights with decay, 1) weights no decay, 2) biases no decay
    g = [], [], []  # optimizer parameter groups
    bn = tuple(v for k, v in nn.__dict__.items() if 'Norm' in k)  # normalization layers, i.e. BatchNorm2d()
    for v in model.cells():
        print(model.cells())
        for p_name, p in v.parameters_and_names():
            if p_name.endswith('gamma'):
                g[1].append(p)
            if p_name.endswith('beta'):
                g[2].append(p)
            if p_name.endswith('bias'):  # bias (no decay)
                g[2].append(p)
            if p_name.endswith('weight'):
                if isinstance(v, bn):  # weight (no decay)
                   g[1].append(p)
                else:
                   g[0].append(p)  # weight (with decay)
    if name == 'Adam':
        optimizer = nn.Adam(g[2], learning_rate=lr, beta1=momentum, beta2=0.999)  # adjust beta1 to momentum
    elif name == 'AdamW':
        optimizer = nn.AdamWeightDecay(g[2], learning_rate=lr, beta1=momentum, beta2=0.999, weight_decay=0.0)
    elif name == 'RMSProp':
        optimizer = nn.RMSProp(g[2], learning_rate=lr, momentum=momentum)
    elif name == 'SGD':
        optimizer = nn.SGD(g[2], learning_rate=lr, momentum=momentum, nesterov=True)
    else:
        raise NotImplementedError(f'Optimizer {name} not implemented.')

optimizer.add_param_group({'params': g[0], 'weight_decay': decay})  # add g0 with weight_decay
    optimizer.add_param_group({'params': g[1], 'weight_decay': 0.0})  # add g1 (BatchNorm2d weights)
    LOGGER.info(f"{colorstr('optimizer:')} {type(optimizer).__name__}(lr={lr}) with parameter groups "
                f"{len(g[1])} weight(decay=0.0), {len(g[0])} weight(decay={decay}), {len(g[2])} bias")
    return optimizer

MindSpore/models

内容风险标识

评论 (22)

MindSpore/models .gitee-modal { width: 500px !important; }

内容风险标识