tf.where(Tensor) --报错：ValueError: Shapes must be equal rank, but are 0 and 3 for ‘ ’

最新推荐文章于 2022-07-31 21:14:42 发布

砸吧砸吧

最新推荐文章于 2022-07-31 21:14:42 发布

阅读量2k

点赞数

文章标签： python tensorflow

本文链接：https://blog.csdn.net/m0_49629753/article/details/121047532

版权

出错代码：

mask_windows = window_partition(mask_array, self.window_size)
            mask_windows = tf.reshape(mask_windows, shape=[-1, self.window_size * self.window_size])
            attn_mask = tf.expand_dims(mask_windows, axis=1) - tf.expand_dims(mask_windows, axis=2)
            attn_mask = tf.cast(attn_mask, tf.float32)
            attn_mask = tf.where(attn_mask != 0, -100.0, attn_mask)
            attn_mask = tf.where(attn_mask == 0, 0.0, attn_mask)

报错：

ValueError: Shapes must be equal rank, but are 0 and 3 for 'generator_swin_unet/swin_transformer_block_1/Select' (op: 'Select') with input shapes: [], [], [196,16,16].

解决方案：

 mask_windows = window_partition(mask_array, self.window_size)
            mask_windows = tf.reshape(mask_windows, shape=[-1, self.window_size * self.window_size])
            attn_mask = tf.expand_dims(mask_windows, axis=1) - tf.expand_dims(mask_windows, axis=2)
#修改shape到相同
            non_zero = tf.fill(tf.shape(attn_mask), -100.0)
            zero = tf.fill(tf.shape(attn_mask), -0.0)  
            attn_mask = tf.where(tf.equal(attn_mask, 0), zero, non_zero)
            # attn_mask = tf.where(attn_mask != 0, non_zero, attn_mask)
            # attn_mask = tf.where(attn_mask == 0, zero, attn_mask)