Import "tensorflow.keras" could not be resolved

[otavio-silva]

Environment data

• Language Server version: 2022.12.20
• OS and version: win32 x64
• Python version (and distribution if applicable, e.g. Anaconda):
• python.analysis.indexing: true
• python.analysis.typeCheckingMode: basic

Code Snippet

import tensorflow as tf
from tensorflow.keras import activations, initializers, layers, losses, metrics, models, optimizers, regularizers

Repro Steps

1. Open VSCode
2. Create somename.py
3. Type code snippet

Expected behavior

Imports should be resolved

Actual behavior

Import "tensorflow.keras" could not be resolvedPylance[reportMissingImports](https://github.com/microsoft/pyright/blob/main/docs/configuration.md#reportMissingImports)

Logs

pylance.log

[rchiodo]

I think tensorflow.keras is a special case. Its location is determined dynamically as far as I can tell, so pylance cannot figure out where it is.

The tensorflow.keras module is found here for me:

.venv/lib/python3.10/site-packages/keras/

whereas pylance would look for it here:

.venv/lib/python3.10/site-packages/tensorflow/keras

@bschnurr has advice on how to get pylance to see where it is:

#3387 (comment)

[otavio-silva]

@rchiodo the comment you linked does nothing, it says that anything name under the tf.keras is of type Any and does not gives suggestions, see:

[bschnurr]

whats your version of tensorflow? try the latest

[otavio-silva]

I installed version 2.10 using mambaforge on WSL2 Ubuntu, it works now.
And it seems that TensorFlow is going for Linux-only, this is the way then.

[AnuraagaNath]

Try keras only without using tensorflow before it

[mnothman]

With keras3 there is a new way to import, the documentation has examples for layers, models, optimizers, applications
https://keras.io/api/

Just do below with no other imports:

import tensorflow as tf
from tensorflow import keras

For Dense you would do

keras.layers.Dense

Example:

x = keras.layers.Dense(128, activation='relu')(x)