Skip to content

feat: removing hard limits on learning rate #3241

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

feat: removing hard limits on learning rate #3241

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 11 additions & 4 deletions unsloth/models/rl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -545,10 +545,17 @@ def _patch_trl_rl_trainers(trainer_file = "grpo_trainer"):
# Warn on too large or too small learning rate
if " learning_rate" in call_args:
learning_rate_check = \
"if learning_rate < 1e-7: raise FloatingPointError(f'Unsloth: Your learning rate of `{learning_rate}` is too small and less than 1e-7! "\
"Consider increasing it, otherwise gradient updates will be close to 0!')\n"\
"if learning_rate > 1: raise OverflowError(f'Unsloth: Your learning rate of `{learning_rate}` is way too larger > 1! "\
"Consider decreasing it to 1e-1, otherwise gradient updates will explode!')\n"
"use_strict_mode = os.environ.get('UNSLOTH_USE_STRICT_MODE', '1') == '1'\n"\
"lower_limit_msg = f'Unsloth: Your learning rate of `{learning_rate}` is too small and less than 1e-7! "\
"Consider increasing it, otherwise gradient updates will be close to 0!'\n"\
"upper_limit_msg = f'Unsloth: Your learning rate of `{learning_rate}` is way too and more than 1! "\
"Consider decreasing it to 1e-1, otherwise gradient updates will explode!'\n"\
"if learning_rate < 1e-7:\n"\
"if use_strict_mode: raise FloatingPointError(lower_limit_msg)\n"\
"else: print(lower_limit_msg)\n"\
"if learning_rate > 1:\n"\
"if use_strict_mode: raise OverflowError(upper_limit_msg)\n"\
"else: print(upper_limit_msg)\n"\
extra_args += learning_rate_check
pass

Expand Down