-
Notifications
You must be signed in to change notification settings - Fork 4.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Removing TensorFlow Trainers #4707
Conversation
We can remove the upper limit on numpy ml-agents/ml-agents-envs/setup.py Line 51 in 3c07161
Which was added to keep TF happy in #4274. |
We should remove the constraint files and their usage ml-agents/.github/workflows/pytest.yml Line 23 in 3c07161
which are used to install various versions of tensorflow during CI. |
Can avoid installing tensorflow in the yamato workflow: ml-agents/ml-agents/tests/yamato/yamato_utils.py Lines 118 to 119 in 65644d1
|
"default framework, and will be removed in the next release.", | ||
) | ||
argparser.add_argument( | ||
"--tensorflow", | ||
default=False, | ||
action=DetectDefaultStoreTrue, | ||
action=RaiseDeprecationWarning, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should we remove this one altogether? (Deprecated) usually means that it still works but will be removed in the future, whereas this flag won't do anything
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I see 3 options:
- Remove --tensorflow and --torch (I think users will complain that a command line that used to work does not anymore)
- Raise a specific error when --tensorflow is used to signal to the user that the argument has been removed
- Only raise a warning when used (the warning can say "feature removed" instead of "feature deprecated"
I have a preference for option 2 but not a strong one. What do you think?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think keeping it around with a warning for one more release is good practice. For the wording, "tensorflow" is "removed", "--tensorflow" and "--torch" are deprecated and have no effect.
@@ -11,6 +11,7 @@ and this project adheres to | |||
### Major Changes | |||
#### com.unity.ml-agents (C#) | |||
#### ml-agents / ml-agents-envs / gym-unity (Python) | |||
- TensorFlow trainers have been removed, please use the Torch trainers instead. (#4707) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This will have to move to move to the 'Unreleased' section after release 11 (unless we are aiming to get this in for the release)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, let's merge after R11 split
Proposed change(s)
Useful links (Github issues, JIRA tickets, ML-Agents forum threads etc.)
Types of change(s)
Checklist
Other comments