If you docker run without attaching a tty, and only call bash, then bash finds nothing to do, and it exits. That’s because by default, a container is non-interactive, and a shell that runs in non-interactive mode expects a script to run. Absent that, it will exit.
To run a disposable new container, you can simply attach a tty and standard input:
docker run --rm -it --entrypoint bash <image-name-or-id>
Or to prevent the above container from being disposed, run it without --rm.
Or to enter a running container, use exec instead:
docker exec -it <container-name-or-id> bash
In comments you asked
Do you know what is the difference between this and
docker run -it --entrypoint bash docker/whalesay?
In the two commands above, you are specifying bash as the CMD. In this command, you are specifying bash as the ENTRYPOINT.
Every container is run using a combination of ENTRYPOINT and CMD. If you (or the image) does not specify ENTRYPOINT, the default entrypoint is /bin/sh -c.
So in the earlier two commands, if you run bash as the CMD, and the default ENTRYPOINT is used, then the container will be run using
/bin/sh -c bash
If you specify --entrypoint bash, then instead it runs
bash <command>
Where <command> is the CMD specified in the image (if any is specified).