Getting pip dependency error following TFX tutorial

Question

I'm trying to follow the Tensorflow Extended tutorial from the official website: https://www.tensorflow.org/tfx/tutorials/tfx/components_keras

When I try executing the line

pip install -q -U --use-feature=2020-resolver tfx

I get the following error:

I haven't been able to find anything online - any help is appreciated :)

score 0 · Answer 1 · answered Feb 11 '21 at 02:19

0

The problem is due to dependency mismatch. Use virtual environment to get rid of such kind of problems.

#Create virtual environment named env
python -m venv env

#Activate the environment
source env/bin/activate

#Now install packages you need using pip
pip install -q -U --use-feature=2020-resolver tfx

These instructions are for Linux OS. If you are using windows let me know. Will help you with that.

answered Feb 11 '21 at 02:19

Dipak Tandel

129
1
2
8

Is this helpful to you? – Dipak Tandel Feb 12 '21 at 04:42
Can you upvote the solution if it's helpful? – Dipak Tandel Feb 20 '21 at 05:23

score 0 · Answer 2 · answered Mar 25 '21 at 12:25

0

Please replace the --use-feature flag with --use-deprecated=legacy-resolver to work around the dependency error.

answered Mar 25 '21 at 12:25

score 0 · Answer 3 · answered Apr 01 '21 at 20:36

There is a TFX tutorial recently published by Adaltas which might help you to get your environment up and running. It is based on Miniconda: https://www.adaltas.com/en/2021/03/05/tfx-overview/. Here is an extract:

$ conda --version
conda 4.9.1
$ conda create --name env_tfx python=3.7
$ conda activate env_tfx
(env_tfx)$ pip install tensorflow==2.3.0  
(env_tfx)$ pip install tfx==0.24.1
(env_tfx)$ pip install notebook

Getting pip dependency error following TFX tutorial

3 Answers3