Hardik

Hi, which version of TFF are you using? I tried running the command with the latest version, `0.26.0`, and wasn't able to reproduce your failure.

I was able to import tff and use it problem-free with Python 3.9, ``` $ python3 --version Python 3.9.12 $ python3 -c "import tensorflow_federated as tff ; print(tff.federated_computation(lambda: 'Hello, World!')())"...

Hi, the nightly version of `tensorflow-federated` is prone to these types of breakages, but we've updated the [notebook](https://www.tensorflow.org/federated/tutorials/federated_learning_for_image_classification) to install the latest [stable release](https://github.com/tensorflow/federated/releases/tag/v0.20.0) instead. Can you try using that...

Hi, it seems that pip is backtracking through `tf_nightly` versions to try to resolve a conflicting dependency that's already installed. Have you tried installing the latest official release of [`tensorflow-federated`](https://github.com/tensorflow/federated/releases/tag/v0.20.0)...

Hi, you should be able to pull `gcr.io/tensorflow-federated/remote-executor-service:0.20.0`.

Makes sense. Although, what would be the use case for future pipelines to write to the directory of a pipeline?

After some thought, I believe it's only Python 2.7 that requires encodings beyond ASCII to be specified...bite the bullet and switch over to Python 3?

There are version incompatibilities between TFF and Python 3.10, across all versions of TFF. But installing `tensorflow-federated==0.36.0` (the latest version) in an environment with Python 3.9 should work.