Apple M1 Max运行TensorFlow报错的排查及解决方法

环境配置 #

• 设备：搭载Apple M1 MAX芯片的Mac
• 系统：Monterey 12.6.1
• 运行环境：Anaconda，Python 3.9.12
• TensorFlow相关包版本：

python --version
Python 3.9.12

conda list | grep tensorflow
tensorflow-deps           2.8.0                         0    apple
tensorflow-estimator      2.10.0                   pypi_0    pypi
tensorflow-macos          2.10.0                   pypi_0    pypi
tensorflow-metal          0.6.0                    pypi_0    pypi

运行代码 #

from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import Adam
from tensorflow.keras import layers

model = Sequential([layers.Input((3, 1)),
                    layers.LSTM(64),
                    layers.Dense(32, activation='relu'),
                    layers.Dense(32, activation='relu'),
                    layers.Dense(1)])

model.compile(loss='mse', 
              optimizer=Adam(learning_rate=0.001),
              metrics=['mean_absolute_error'])

model.fit(X_train, y_train, validation_data=(X_val, y_val), epochs=100)

报错信息 #

核心错误 #

RuntimeError: module compiled against API version 0x10 but this version of numpy is 0xe

回溯信息 #

TypeError                                 Traceback (most recent call last)
Input In [14], in <cell line: 1>()
----> 1 from tensorflow.keras.models import Sequential
      2 from tensorflow.keras.optimizers import Adam
      3 from tensorflow.keras import layers

该报错仅在M1 MAX芯片的Mac上出现，2020款M1芯片的Mac mini（Monterey 12.6系统）及Win11系统中运行相同代码无异常，怀疑是环境或芯片适配问题。目标是成功构建模型并完成训练数据拟合。

解决步骤 #

通过以下操作解决了依赖冲突及numpy版本不匹配问题：

• 卸载keras：pip uninstall keras
• 卸载keras-preprocessing：pip uninstall keras-preprocessing
• 卸载tensorboard：pip uninstall tensorboard
• 升级numpy：pip install --upgrade numpy
• 若升级numpy失败，先卸载再重新安装：pip uninstall numpy && pip install numpy
• 重新安装tensorflow-macos：python -m pip install tensorflow-macos

内容的提问来源于stack exchange，提问作者Egypt CCI