我不断收到 TensorFlow ValueError，但我不知道是什么原因导致的

Question

所以我安装了所有东西来在 Visual Studio 代码中运行 jupyter 笔记本，并完全更新了 TensorFlow 和 Python。但是，当我运行其中一个内核时，我不断遇到错误。具体来说，

File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\keras\src\utils\traceback_utils.py:122, in filter_traceback.<locals>.error_handler(*args, **kwargs)
    119     filtered_tb = _process_traceback_frames(e.__traceback__)
    120     # To get the full stack trace, call:
    121     # `keras.config.disable_traceback_filtering()`
--> 122     raise e.with_traceback(filtered_tb) from None
    123 finally:
    124     del filtered_tb

File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\keras\src\trainers\data_adapters\__init__.py:113, in get_data_adapter(x, y, sample_weight, batch_size, steps_per_epoch, shuffle, class_weight)
    105     return GeneratorDataAdapter(x)
    106     # TODO: should we warn or not?
    107     # warnings.warn(
    108     #     "`shuffle=True` was passed, but will be ignored since the "
   (...)
    111     # )
    112 else:
--> 113     raise ValueError(f"Unrecognized data type: x={x} (of type {type(x)})

最后显示 ValueError: 无法识别的数据类型 x=[10.0] (of type )

有什么办法可以解决这个问题吗？

这是我当前的代码：

import tensorflow as tf
import numpy as np
from tensorflow import keras
model = tf.keras.Sequential([keras.layers.Dense(units=1, input_shape=[1])])
model.compile(optimizer='sgd', loss='mean_squared_error')
xs = np.array([-1.0,  0.0, 1.0, 2.0, 3.0, 4.0], dtype=float)
ys = np.array([-3.0, -1.0, 1.0, 3.0, 5.0, 7.0], dtype=float)
model.fit(xs, ys, epochs=500)
print(model.predict([10.0])) #This is where the error is

这就是值错误的地方

if array_data_adapter.can_convert_arrays((x, y, sample_weight)):
        return ArrayDataAdapter(
            x,
            y,
            sample_weight=sample_weight,
            class_weight=class_weight,
            shuffle=shuffle,
            batch_size=batch_size,
            steps=steps_per_epoch,
        )
    elif is_tf_dataset(x):
        # Unsupported args: y, sample_weight, shuffle
        if y is not None:
            raise_unsupported_arg("y", "the targets", "tf.data.Dataset")
        if sample_weight is not None:
            raise_unsupported_arg(
                "sample_weights", "the sample weights", "tf.data.Dataset"
            )
        return TFDatasetAdapter(
            x, class_weight=class_weight, distribution=distribution
        )
        # TODO: should we warn or not?
        # warnings.warn(
        #     "`shuffle=True` was passed, but will be ignored since the "
        #     "data `x` was provided as a tf.data.Dataset. The Dataset is "
        #     "expected to already be shuffled "
        #     "(via `.shuffle(tf.data.AUTOTUNE)`)"
        # )
    elif isinstance(x, py_dataset_adapter.PyDataset):
        if y is not None:
            raise_unsupported_arg("y", "the targets", "PyDataset")
        if sample_weight is not None:
            raise_unsupported_arg(
                "sample_weights", "the sample weights", "PyDataset"
            )
        return PyDatasetAdapter(x, class_weight=class_weight, shuffle=shuffle)
    elif is_torch_dataloader(x):
        if y is not None:
            raise_unsupported_arg("y", "the targets", "torch DataLoader")
        if sample_weight is not None:
            raise_unsupported_arg(
                "sample_weights", "the sample weights", "torch DataLoader"
            )
        if class_weight is not None:
            raise ValueError(
                "Argument `class_weight` is not supported for torch "
                f"DataLoader inputs. Received: class_weight={class_weight}"
            )
        return TorchDataLoaderAdapter(x)
        # TODO: should we warn or not?
        # warnings.warn(
        #     "`shuffle=True` was passed, but will be ignored since the "
        #     "data `x` was provided as a torch DataLoader. The DataLoader "
        #     "is expected to already be shuffled."
        # )
    elif isinstance(x, types.GeneratorType):
        if y is not None:
            raise_unsupported_arg("y", "the targets", "PyDataset")
        if sample_weight is not None:
            raise_unsupported_arg(
                "sample_weights", "the sample weights", "PyDataset"
            )
        if class_weight is not None:
            raise ValueError(
                "Argument `class_weight` is not supported for Python "
                f"generator inputs. Received: class_weight={class_weight}"
            )
        return GeneratorDataAdapter(x)
        # TODO: should we warn or not?
        # warnings.warn(
        #     "`shuffle=True` was passed, but will be ignored since the "
        #     "data `x` was provided as a generator. The generator "
        #     "is expected to yield already-shuffled data."
        # )
    else:
        raise ValueError(f"Unrecognized data type: x={x} (of type {type(x)})")

这是突出显示的另一段代码

def filter_traceback(fn):
    """Filter out Keras-internal traceback frames in exceptions raised by fn."""

    @wraps(fn)
    def error_handler(*args, **kwargs):
        if not is_traceback_filtering_enabled():
            return fn(*args, **kwargs)

        filtered_tb = None
        try:
            return fn(*args, **kwargs)
        except Exception as e:
            filtered_tb = _process_traceback_frames(e.__traceback__)
            # To get the full stack trace, call:
            # `keras.config.disable_traceback_filtering()`
            raise e.with_traceback(filtered_tb) from None
        finally:
            del filtered_tb

Answer 1

您可以直接在 Tensorflow 中使用

list

对象。改变

print(model.predict([10.0]))

到

pred = np.array([10.0]).reshape(1, -1)
print(model.predict(pred))

或者更好

pred = tf.reshape(tf.convert_to_tensor([10.0]), (1, -1))
print(model.predict(pred))

将列表

[10.0]

转换为 numpy 数组（如

model.fit(..)

调用的数组），或直接转换为 Tensorflow 张量。如果只有一个样本要预测，则需要向形状添加批次维度 1，因为 Tensorflow 期望输入为形状（批次、特征），即使批次只有 1 个元素。