-
Notifications
You must be signed in to change notification settings - Fork 172
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
Retrieve schedules from get_schedule_class() #595
base: gh/H-Huang/17/base
Are you sure you want to change the base?
Conversation
[ghstack-poisoned]
ghstack-source-id: 8abfc98eb138e904749b632dc4250ee9da79314f Pull Request resolved: #595
torchtitan/config_manager.py
Outdated
@@ -299,14 +299,14 @@ def __init__(self): | |||
self.parser.add_argument( | |||
"--experimental.pipeline_parallel_schedule", | |||
type=str, | |||
choices=["1f1b", "gpipe", "interleaved_1f1b", "flexible_interleaved_1f1b"], | |||
choices=["1f1b", "gpipe", "Interleaved1F1B", "FlexibleInterleaved1F1B"], |
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.
any reason we leave 1f1b
and gpipe
in lower cases?
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.
The get_schedule()
method is case insensitive, the main change was i needed to remove the "_" in the other schedules. But I can change these as well for consistency.
Removes glue code and allow users to pass in schedule arg as a string without updating titan. Depends on pytorch/pytorch#137103 [ghstack-poisoned]
@@ -299,14 +299,14 @@ def __init__(self): | |||
self.parser.add_argument( | |||
"--experimental.pipeline_parallel_schedule", | |||
type=str, | |||
choices=["1f1b", "gpipe", "interleaved_1f1b", "flexible_interleaved_1f1b"], | |||
default="1f1b", | |||
choices=["1F1B", "Gpipe", "Interleaved1F1B", "FlexibleInterleaved1F1B"], |
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.
very tiny nit: GPipe
instead of Gpipe
Stack from ghstack (oldest at bottom):
Removes glue code and allow users to pass in schedule arg as a string without updating titan.
Depends on pytorch/pytorch#137103