Could not find a version that satisfies the requirement tensorflow~=1.15.0

I solved the issue by switching to Docker builds.

#1 Added heroku.yml to the root of my project:

build:
  docker:
    web: Dockerfile
run:
  web: python app.py

#2 pip upgrade happens in the Dockerfile:

FROM python:3.7-slim
...
RUN pip install --upgrade pip
RUN pip install --no-cache-dir -r requirements.txt
...

#3 Don’t forget to change the type of your Heroku app:

heroku stack:set container --app YOUR_APP_NAME

This way I was able to control both the Python and the pip version. Prior to moving to the Docker deployments I tried upgrading Python to the latest stack in Heroku (through runtime.txt), but to no avail: pip won’t get upgraded beyond 9.0.x.

1 Like

Below step will help pip3 install tensorflow==1.5 pip3 install rasa-x --extra-index-url https://pypi.rasa.com/simple --ignore-installed tensorflow

Looks like the problem is with Python 3.8. Use Python 3.7 instead. Steps I took to solve this.

  1. Created a python 3.7 environment with conda
  2. Installed rasa using pip install rasa within the environment.

Worked for me.