[grpo]Tool rl: add reward func for ToolRL #4694
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
hjh0119
reviewed
Jun 26, 2025
examples/train/grpo/plugin/plugin.py
Outdated
| max_possible_reward = self.format_max_possible | ||
| min_possible_reward = self.format_min_possible | ||
| if str(os.getenv("MAX1STEP30MAX3", 0)) == "1": | ||
| print("MAX1STEP30MAX3 is set to 1, so max 1 -> 30 steps -> max 3") |
There was a problem hiding this comment.
Remove print statements; otherwise, there will be a lot of print messages in the logger.
There was a problem hiding this comment.
All print statements have been removed.
examples/train/grpo/plugin/plugin.py
Outdated
|
|
||
| # schedule reward | ||
| if str(os.getenv("SCHEDULEREWARD", 0)) == "1": | ||
| print("SCHEDULEREWARD is set to 1, so schedule reward is used") |
examples/train/grpo/plugin/plugin.py
Outdated
Comment on lines
483
to
485
| print("\n======= Answer ======= ") | ||
| print(solution[0]) | ||
| print("\n======= Responses ======= ") |
examples/train/grpo/plugin/plugin.py
Outdated
| print(solution[0]) | ||
| print("\n======= Responses ======= ") | ||
| for idx, response in enumerate(responses): | ||
| print(f"*** Response {idx+1}***\n{response}") |
examples/train/grpo/plugin/plugin.py
Outdated
Comment on lines
510
to
512
| print("\n======= Reward for <format> =======") | ||
| print("Reward function for <format> is called ...") | ||
| print(rewards) |
examples/train/grpo/plugin/plugin.py
Outdated
Comment on lines
586
to
587
| print("Max possible score:", "Exact Match!") | ||
| print("Score:", max_possible_reward) |
examples/train/grpo/plugin/plugin.py
Outdated
| return max_possible_reward | ||
|
|
||
| if os.getenv("COARSEREWARD", 0) == "1": | ||
| print("COARSEREWARD is set to 1, so coarse reward is used") |
examples/train/grpo/plugin/plugin.py
Outdated
| gt_params = gt_tool["parameters"] | ||
|
|
||
| if str(os.getenv("INTERMEDIATEREWARD", 0)) == "1": | ||
| print("INTERMEDIATEREWARD is set to 1, so local max possible is changed") |
examples/train/grpo/plugin/plugin.py
Outdated
Comment on lines
647
to
649
| print() | ||
| print("Max possible score:", local_max_possible) | ||
| print("Score:", score) |
examples/train/grpo/plugin/plugin.py
Outdated
Comment on lines
712
to
714
| print("\n======= Reward for <tool call> =======") | ||
| print("Reward function for <tool call> correctness is called ...") | ||
| print(rewards) |
|
plz pass the lint test |
Lint test has been fixed |
hjh0119
approved these changes
Jun 27, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR type
PR information
add reward func for ToolRL(https://arxiv.org/abs/2504.13958).
Experiment results
Paste your experiment result here(if needed).