Skip to content

Linear LR scheduler #150

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.

Sign up for GitHub

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

Jump to bottom
Merged
merged 2 commits into from
Dec 25, 2022
Merged

Linear LR scheduler #150

merged 2 commits into from
Dec 25, 2022

Conversation

leshanbog
Copy link
Contributor

@leshanbog leshanbog commented Dec 24, 2022 •
edited
Loading

In this PR:

  • Linear LR scheduler is added. It was used in InstructGPT. From the paper:

image

  • Tested with config:
scheduler:
  name: "linear"
  kwargs:
    start_factor: 0.1
    end_factor: 1.0
    total_iters: 64

image

  • Some unused code is removed

jon-tow
jon-tow approved these changes Dec 25, 2022
View reviewed changes
Copy link
Collaborator

@jon-tow jon-tow left a comment •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for another great PR :)

Note: In the far future we may want to consider using StrEnum for Python >= 3.11 as 3.11 introduces a breaking change in f-string handling of Enum str mixin accesses (f"{Enum.FOO}" does not return the value string of FOO).

trlx/utils/__init__.py
Comment on lines -30 to -42
def flatten(L: Iterable[Iterable[Any]]) -> Iterable[Any]:
"""
Flatten a list of lists into a single list (i.e. [[1, 2], [3, 4]] -> [1,2,3,4])
"""
return list(reduce(lambda acc, x: acc + x, L, []))


def chunk(L: Iterable[Any], chunk_size: int) -> List[Iterable[Any]]:
"""
Chunk iterable into list of iterables of given chunk size
"""
return [L[i : i + chunk_size] for i in range(0, len(L), chunk_size)]

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

NIce cleanup 👍

trlx/utils/__init__.py


class OptimizerNames(Enum):
class OptimizerName(str, Enum):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good idea with the str mixin.

@jon-tow jon-tow merged commit 40126e8 into CarperAI:main Dec 25, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@jon-tow jon-tow jon-tow approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@leshanbog @jon-tow