Created by: lxk767363331

!python image_classification/train.py \
        --data_dir=breast/ \
        --total_images=6356 \
        --class_dim=8 \
        --validate=True \
        --model=HRNet_W32_C \
        --batch_size=16 \
        --lr_strategy=cosine_decay \
        --image_shape 3 700 460 \
        --lr=0.001 \
        --num_epochs=200 \
        --model_save_dir=output/ \
        --l2_decay=7e-5 \
        --use_mixup=True \
        --use_label_smoothing=True \
        --label_smoothing_epsilon=0.1 \
        --pretrained_model=best/HRNet_W32_C  

使用这个指令进行训练时 增加 --image_shape 3 700 460 \ 报错结果如下： ValueError: The feeded Variable 'feed_image' should have dimensions = 4, shape = (-1, 3, 700, 460), but received feeded shape [16, 3, 700, 700] 请问在代码中如何设置此项，我在reader中没有找到？ 还有一个问题
我尝试不适用预训练模型 分别使用 --image_shape 3 700 460 / --image_shape 3 460 460 均报错，可否有解释，如何更改？ -image_shape 3 460 460 报错结果如下： Error Message Summary:

Error: ShapeError: broadcast dimension mismatch. Operands could not be broadcast together with the shape of X = [16, 32, 115, 115] and the shape of Y = [16, 32, 116, 116]. Received [115] in X is not equal to [116] in Y [Hint: Expected y_dims[i] == 1, but received y_dims[i]:116 != 1:1.] at (/paddle/paddle/fluid/operators/elementwise/elementwise_op_function.h:79) [operator < elementwise_add > error]