[BREAKING] config: set the default value of actor.entropy_coeff to 0 #1770
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
vermouth1992
approved these changes
May 30, 2025
yzlnew
pushed a commit
to yzlnew/verl
that referenced
this pull request
Jun 4, 2025
…olcengine#1770) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? entropy_coeff shall be set carefully during RL. When enabled, inappropriate coefficient may case training to collapse. You can see more empirical experiments from Skywork Open Reasoner 1 Technical Report (https://arxiv.org/pdf/2505.22312). In this PR, the default value of entropy_coeff is set to 0. This is a breaking change that may affect your experiment, although majority of verl example scripts set it to 0 manually already. We let most example script just pick up the default value of 0 for entropy_coeff. For a few documentation page where the reference model performance and commands are provided, we modify the doc so that the experiment result is consistent with the config setup. ### Usage Example To enable entropy loss coefficient, use ```bash actor_rollout_ref.actor.entropy_coeff=0.001 # or other values ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [x] Add `[BREAKING]` to the PR title if it breaks any API. - [x] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if necessary.
wwwjn
pushed a commit
to wwwjn/verl
that referenced
this pull request
Jun 10, 2025
…olcengine#1770) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? entropy_coeff shall be set carefully during RL. When enabled, inappropriate coefficient may case training to collapse. You can see more empirical experiments from Skywork Open Reasoner 1 Technical Report (https://arxiv.org/pdf/2505.22312). In this PR, the default value of entropy_coeff is set to 0. This is a breaking change that may affect your experiment, although majority of verl example scripts set it to 0 manually already. We let most example script just pick up the default value of 0 for entropy_coeff. For a few documentation page where the reference model performance and commands are provided, we modify the doc so that the experiment result is consistent with the config setup. ### Usage Example To enable entropy loss coefficient, use ```bash actor_rollout_ref.actor.entropy_coeff=0.001 # or other values ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [x] Add `[BREAKING]` to the PR title if it breaks any API. - [x] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if necessary.
chenjiaoAngel
added a commit
to chenjiaoAngel/verl
that referenced
this pull request
Nov 14, 2025
…olcengine#1770) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? entropy_coeff shall be set carefully during RL. When enabled, inappropriate coefficient may case training to collapse. You can see more empirical experiments from Skywork Open Reasoner 1 Technical Report (https://arxiv.org/pdf/2505.22312). In this PR, the default value of entropy_coeff is set to 0. This is a breaking change that may affect your experiment, although majority of verl example scripts set it to 0 manually already. We let most example script just pick up the default value of 0 for entropy_coeff. For a few documentation page where the reference model performance and commands are provided, we modify the doc so that the experiment result is consistent with the config setup. ### Usage Example To enable entropy loss coefficient, use ```bash actor_rollout_ref.actor.entropy_coeff=0.001 # or other values ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [x] Add `[BREAKING]` to the PR title if it breaks any API. - [x] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if necessary.
TimurTaepov
pushed a commit
to giorgossideris/verl
that referenced
this pull request
Dec 20, 2025
…olcengine#1770) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? entropy_coeff shall be set carefully during RL. When enabled, inappropriate coefficient may case training to collapse. You can see more empirical experiments from Skywork Open Reasoner 1 Technical Report (https://arxiv.org/pdf/2505.22312). In this PR, the default value of entropy_coeff is set to 0. This is a breaking change that may affect your experiment, although majority of verl example scripts set it to 0 manually already. We let most example script just pick up the default value of 0 for entropy_coeff. For a few documentation page where the reference model performance and commands are provided, we modify the doc so that the experiment result is consistent with the config setup. ### Usage Example To enable entropy loss coefficient, use ```bash actor_rollout_ref.actor.entropy_coeff=0.001 # or other values ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [x] Add `[BREAKING]` to the PR title if it breaks any API. - [x] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if necessary.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Checklist Before Starting
What does this PR do?
entropy_coeff shall be set carefully during RL. When enabled, inappropriate coefficient may case training to collapse. You can see more empirical experiments from Skywork Open Reasoner 1 Technical Report (https://arxiv.org/pdf/2505.22312).
In this PR, the default value of entropy_coeff is set to 0. This is a breaking change that may affect your experiment, although majority of verl example scripts set it to 0 manually already.
We let most example script just pick up the default value of 0 for entropy_coeff. For a few documentation page where the reference model performance and commands are provided, we modify the doc so that the experiment result is consistent with the config setup.
Usage Example
To enable entropy loss coefficient, use
actor_rollout_ref.actor.entropy_coeff=0.001 # or other valuesTest
Additional Info.
Checklist Before Submitting
[BREAKING]to the PR title if it breaks any API.