警告：tensorflow：过去 11 次调用中的 11 次触发 tf.function 回溯

Question

有人知道这个错误的原因吗？

WARNING:tensorflow:No training configuration found in the save file, so the model was *not* compiled. Compile it manually.
WARNING:tensorflow:11 out of the last 11 calls to <function Model.make_predict_function.<locals>.predict_function at 0x000001F9D1C05EE0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.
WARNING:tensorflow:11 out of the last 11 calls to <function Model.make_predict_function.<locals>.predict_function at 0x000001F9D5604670> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.
C:\Users\User\anaconda3\lib\site-packages\sklearn\cluster\_kmeans.py:973: FutureWarning: 'n_jobs' was deprecated in version 0.23 and will be removed in 0.25.
  warnings.warn("'n_jobs' was deprecated in version 0.23 and will be"

Answer 1

{TLDR} 尝试将 model.predict(x) 替换为 model(x)

我的解决方案

我也遇到了警告问题：

WARNING:tensorflow:11 out of the last 11 calls to <function Model.make_predict_function.<locals>.predict_function at 0x000001F9D1C05EE0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.

我能够通过直接使用 model(x)

替换 model.predict(x) 来解决它

---

我遇到问题的背景信息

我正在预测时间序列，并在每个新采样时间将模型的最后一层拟合到最新数据。所以我

1. 生成并拟合基础模型并冻结所有层 + 放置顶层
2. 适应新数据并在循环内预测

我尝试使用警告和@TFer2 中建议的签名来实现自定义预测函数。然而，这产生了错误

RuntimeError: Detected a call to `Model.predict` inside a `tf.function`. `Model.predict is a high-level endpoint that manages its own `tf.function`. Please move the call to `Model.predict` outside of all enclosing `tf.function`s. Note that you can call a `Model` directly on `Tensor`s inside a `tf.function` like: `model(x)`.

有了这个错误，我就可以解决问题了。

Answer 2

如果您调用具有相同参数类型的函数，tensorflow 将重用先前跟踪的图，否则将创建新图。

函数通过计算cache key from an input's args and kwargs来确定是否重用被跟踪的具体函数：

• 为tf.Tensor 参数生成的密钥是它的shape 和type（输入签名）
• 为tf.Variable 参数生成的键是它的id()。
• 为python 原语生成的密钥是它的value。
• 为嵌套的dicts, lists, tuples, namedtuples 和attrs 生成的键是flattened tuple。

回溯可确保 tensorflow 为每组输入生成正确的图表。但是很贵。

您必须避免过度回溯，否则 tensorflow 通常会发出上述警告。

有几种方法可以控制跟踪行为：

• 指定input_signature in tf.function
• 指定 [None] dimension in tf.TensorSpec 以实现跟踪重用的灵活性
• Cast python arguments to Tensors 减少回溯

更多详情可以参考Better performance with tf.function。