Refactor Detailed Balance by hyeok9855 · Pull Request #432 · GFNOrg/torchgfn

hyeok9855 · 2025-11-10T16:35:53Z

I've read the .github/CONTRIBUTING.md file
My code follows the typing guidelines
I've added appropriate tests
I've run pre-commit hooks locally

Description

Refactor detailed balance score calculation
Minor refactoring of several GFlowNet classes

The results from python tutorials/examples/train_hypergrid_simple.py --loss DB are identical to the master branch.

younik

good, thank you @hyeok9855 !

src/gfn/containers/transitions.py

younik · 2025-11-13T17:16:53Z

src/gfn/gflownet/detailed_balance.py

-        scores = preds - targets
+        # Apply forward-looking if applicable
+        if self.forward_looking:
+            import warnings  # type: ignore


logging should have typing

What do you mean?

im not sure I understand either. but why is type: ignore required for an import?

I thought it was required to avoid a Python error, but it turned out to be okay without it. Fixed.

josephdviviano

lgtm! but in the future I think it would be easier for me if you describe in your PR the intention behind the change (e.g., why you are doing a particular refactor) - it's hard for me to see why this was a priority for you, and I think knowing that would help me catch more subtle aspects of the changes.

src/gfn/containers/transitions.py

josephdviviano · 2025-11-19T08:20:24Z

src/gfn/gflownet/detailed_balance.py

            flow of the states.
        forward_looking: Whether to use the forward-looking GFN loss.
-        log_reward_clip_min: If finite, clips log rewards to this value.
-        safe_log_prob_min: If True, uses -1e10 as the minimum log probability value


why did you remove this? it's useful.

log_reward_clip_min is in the PFBasedGFlowNet now

safe_log_prob_min was not used anywhere.

josephdviviano · 2025-11-19T08:27:14Z

src/gfn/gflownet/detailed_balance.py

-        scores = preds - targets
+        # Apply forward-looking if applicable
+        if self.forward_looking:
+            import warnings  # type: ignore


im not sure I understand either. but why is type: ignore required for an import?

src/gfn/gflownet/detailed_balance.py

josephdviviano · 2025-11-19T08:29:00Z

src/gfn/gflownet/base.py

-    ) -> None:
-        """Initializes a TrajectoryBasedGFlowNet instance.
-
-        Args:


I'm a bit confused about how you were able to get rid of this, but seems fine!

It calls super().__init__(...) and does nothing else, so I could remove it.

josephdviviano · 2025-11-19T15:53:24Z

That's funny, we needed the safe log prob min to solve a much earlier issue. I suppose it was removed by someone. (see reply - i was wrong) Joseph (Mobile)

…

On Wed, Nov 19, 2025 at 10:51 Sanghyeok Choi ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In src/gfn/gflownet/detailed_balance.py <#432 (comment)>: > @@ -60,11 +60,9 @@ class DBGFlowNet(PFBasedGFlowNet[Transitions]): logF: A ScalarEstimator or ConditionalScalarEstimator for estimating the log flow of the states. forward_looking: Whether to use the forward-looking GFN loss. - log_reward_clip_min: If finite, clips log rewards to this value. - safe_log_prob_min: If True, uses -1e10 as the minimum log probability value - log_reward_clip_min is in the PFBasedGFlowNet now - safe_log_prob_min was not used anywhere. — Reply to this email directly, view it on GitHub <#432 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA7TL2U3TD4DHOB2EE2CGXL35SGW3AVCNFSM6AAAAACLVXASXWVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTIOBTGQ4TQNBYGU> . You are receiving this because your review was requested.Message ID: ***@***.***>

josephdviviano · 2025-11-19T16:01:06Z

That's funny, we needed the safe log prob min to solve a much earlier issue. I suppose it was removed by someone. Joseph (Mobile)
…
On Wed, Nov 19, 2025 at 10:51 Sanghyeok Choi @.> wrote: @.* commented on this pull request. ------------------------------ In src/gfn/gflownet/detailed_balance.py <#432 (comment)>: > @@ -60,11 +60,9 @@ class DBGFlowNet(PFBasedGFlowNet[Transitions]): logF: A ScalarEstimator or ConditionalScalarEstimator for estimating the log flow of the states. forward_looking: Whether to use the forward-looking GFN loss. - log_reward_clip_min: If finite, clips log rewards to this value. - safe_log_prob_min: If True, uses -1e10 as the minimum log probability value - log_reward_clip_min is in the PFBasedGFlowNet now - safe_log_prob_min was not used anywhere. — Reply to this email directly, view it on GitHub <#432 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA7TL2U3TD4DHOB2EE2CGXL35SGW3AVCNFSM6AAAAACLVXASXWVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTIOBTGQ4TQNBYGU . You are receiving this because your review was requested.Message ID: @.***>

Actually I just looked through the logs and it looks like it was never used anywhere. I guess I forgot to wire it in or something. I wonder why I did that??

hyeok9855 · 2025-11-19T16:02:12Z

lgtm! but in the future I think it would be easier for me if you describe in your PR the intention behind the change (e.g., why you are doing a particular refactor) - it's hard for me to see why this was a priority for you, and I think knowing that would help me catch more subtle aspects of the changes.

@josephdviviano
The work in this PR was originally a part of #431, but I separated it so as not to make each PR too big.

The intention behind these is... I just saw some suboptimal designs in the repo and wanted to improve them.

josephdviviano · 2025-11-19T17:13:02Z

Great work! Thanks for catching that stuff! Joseph (Mobile)

…

On Wed, Nov 19, 2025 at 11:02 Sanghyeok Choi ***@***.***> wrote: *hyeok9855* left a comment (GFNOrg/torchgfn#432) <#432 (comment)> lgtm! but in the future I think it would be easier for me if you describe in your PR the intention behind the change (e.g., why you are doing a particular refactor) - it's hard for me to see why this was a priority for you, and I think knowing that would help me catch more subtle aspects of the changes. @josephdviviano <https://github.com/josephdviviano> The work in this PR was originally a part of #431 <#431>, but I separated it so as not to make each PR too big. The intention behind these is... I just saw some suboptimal designs in the repo and wanted to correct them. — Reply to this email directly, view it on GitHub <#432 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA7TL2TFABD646JDRWHKFDD35SIBXAVCNFSM6AAAAACLVXASXWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTKNJTGUYDAMBWG4> . You are receiving this because you were mentioned.Message ID: ***@***.***>

refactor db

00941d9

hyeok9855 requested review from josephdviviano, saleml and younik November 10, 2025 16:35

hyeok9855 self-assigned this Nov 10, 2025

hyeok9855 added 4 commits November 10, 2025 16:41

add db and fm in train_hypergrid_simple.py

ac108cb

Merge branch 'fix-pytest' into refactor-db

a86889a

fix fl-db

7400b6a

Merge branch 'master' into refactor-db

dec1c9f

younik approved these changes Nov 13, 2025

View reviewed changes

josephdviviano approved these changes Nov 19, 2025

View reviewed changes

remove type: ignore

5f4ce7e

hyeok9855 merged commit a88b715 into master Nov 19, 2025
3 checks passed

Conversation

hyeok9855 commented Nov 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Uh oh!

younik left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

josephdviviano left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

josephdviviano commented Nov 19, 2025 via email • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

josephdviviano commented Nov 19, 2025

Uh oh!

hyeok9855 commented Nov 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

josephdviviano commented Nov 19, 2025 via email

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

hyeok9855 commented Nov 10, 2025 •

edited

Loading

josephdviviano commented Nov 19, 2025 via email •

edited

Loading

hyeok9855 commented Nov 19, 2025 •

edited

Loading