-
Notifications
You must be signed in to change notification settings - Fork 347
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
docs: Release notes for 0.13.8. #1603
Conversation
docs/release-notes.txt
Outdated
simple deployment type). This allows users to explicitly pick a | ||
subnet/availability zone that has GPU instances. | ||
|
||
- Security: Allow the agent to use different addresses for master |
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 don't really follow this or understand why it is useful.
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.
There's an expanded version coming from the release spreadsheet:
Allow the master to use a TLS certificate that is valid for a different name than the agents use to connect to it. This ability is useful in situations where the master is accessed using multiple different addresses (e.g., private and public IP addresses of a cloud instance).
The agent now accepts a ``--security-tls-master-cert-name`` option to override the expected name in the master's TLS certificate. The CLI uses the ``DET_MASTER_CERT_NAME`` environment variable for the same purpose.
docs/release-notes.txt
Outdated
- Deployment: When using ``det-deploy aws``, you can now specify the | ||
subnet that the agents should be launched into (if you are using the | ||
simple deployment type). This allows users to explicitly pick a | ||
subnet/availability zone that has GPU instances. |
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.
Not sure how to actually use this feature -- the docs (how-to/installation/aws.txt
) haven't been updated, we don't give example usage here, and the relationship between subnets and AZs isn't immediately clear to me.
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.
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.
How about:
When using the 'simple' deployment type with 'det deploy aws', you can now use the '--agent-subnet-id' flag to specify which existing subnet to launch agents in. As each subnet is associated with a single availability zone, this allows users to explicitly choose an availability zone that has GPU instances (there is no public information about which availability zones have GPU instances so trial and error is the suggested approach).
I'm on my phone and can't find backticks, but 'det-deploy aws' and '--agent-subnet-id' should be formatted as code.
docs/release-notes.txt
Outdated
the new login endpoint described at | ||
https://docs.determined.ai/latest/rest-api/ . | ||
|
||
- **Breaking Change:** Logging: Start using Fluent Bit for handling trial |
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.
What are the user-visible consequences of this change? @dzhu
e.g., does this result in better scalability, more readable logs, more options for storing logs, ... ?
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.
Honestly, not much by itself (though it does come along with the filtering changes, which I'm going to add a line for separately). Do you think it's worth mentioning here how this is laying the groundwork for future things?
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.
Sure, I don't think that would hurt.
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.
Okay, how about this (appended to what's there now)?
This change is part of an effort to improve the handling of trial logs by increasing scalability and allowing more options for log storage.
docs/release-notes.txt
Outdated
- API: Add ``self.context.configure_fit()`` to the Determined | ||
TFKerasTrial API. Please see :ref:`tf-keras-trial` for detail. | ||
|
||
- Kubernetes: Add option to configure the service type of the Determined |
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.
What is this useful for? Can we give an example of a "service type" you might want to configure? @aaron276h
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.
done
|
||
- API: Add tf.keras DCGAN example that subclasses ``tf.keras.Model``. | ||
|
||
- API: Add ``self.context.configure_fit()`` to the Determined |
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.
What does this actually enable? Can we give an example of what this is useful for? @rb-determined-ai
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.
how about this:
- API: Add ``self.context.configure_fit()`` to the Determined
TFKerasTrial API. Many parameters which would be passed to
``model.fit()``, such as ``class_weight``, ``verbose``, or
``workers``, can now be passed to ``configure_fit()`` and will be
honored by ``TFKerasTrial``. Please see :ref:`tf-keras-trial` for
detail.
- API: Add support for models that subclass ``tf.keras.Model`` when | ||
using the Determined TFKerasTrial API. | ||
|
||
- API: Add tf.keras DCGAN example that subclasses ``tf.keras.Model``. |
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 commit (765275a) did not update docs/examples/index.txt
for the new DCGAN example -- can you fix, @aaron276h ?
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.
Good call, updated this.
docs/release-notes.txt
Outdated
- API: Add ``self.context.wrap_optimizer()`` to the Determined | ||
TFKerasTrial API. Please see :ref:`tf-keras-trial` for detail. | ||
|
||
- API: Add support for models that subclass ``tf.keras.Model`` when |
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.
Can we add a bit more detail here on why this is useful or what you can do with it? Doesn't this qualify as a major feature we should add to the "New Features" list? @aaron276h
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.
done
Co-authored-by: aaron276h <aaron@determined.ai> (cherry picked from commit 80bf3fe)
Co-authored-by: aaron276h <aaron@determined.ai> (cherry picked from commit 80bf3fe)
Description
Test Plan
Commentary (optional)
Checklist
docs/release-notes/
.See Release Note for details.