fix(examples): correct recipe-override precedence asserts in SFT notebook#6049
Open
Hannah14295 wants to merge 1 commit into
Open
fix(examples): correct recipe-override precedence asserts in SFT notebook#6049Hannah14295 wants to merge 1 commit into
Hannah14295 wants to merge 1 commit into
Conversation
Hannah14295
temporarily deployed
to
manual-approval
July 17, 2026 23:22 — with 
GitHub Actions
Inactive
GitHub Actions
Inactive
…book The precedence-verification cell asserted training_config["learning_rate"] and ["num_epochs"], but the resolved Nova recipe exposes neither key, so the notebook fails with KeyError: 'learning_rate'. - The learning_rate override is applied under training_config.optim_config.lr (value 5e-6), not a top-level training_config.learning_rate. - num_epochs is not a Nova recipe key — Nova SFT is step-based (max_steps), so the recipe resolver drops the num_epochs override. Override and assert max_steps (the step-based equivalent of epochs) instead of num_epochs, and assert optim_config.lr. Precedence table updated to match. Verified the overrides take effect (optim_config.lr=5e-6, max_steps=20).
Hannah14295
force-pushed
the
fix/recipe-override-sft-resolved-recipe-keys
branch
from
July 18, 2026 00:10
4c9953c to
0e5df55
Compare
Hannah14295
temporarily deployed
to
manual-approval
July 18, 2026 00:10 — with
GitHub Actions
Inactive
GitHub Actions
Inactive
aviruthen
approved these changes
Jul 18, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
recipe_override_sft_trainer_example asserted on training_config["learning_rate"] and training_config["num_epochs"], but the resolved Nova recipe exposes neither key, so the notebook fails with KeyError: 'learning_rate'.
Assert on training_config["optim_config"]["lr"] and drop the invalid num_epochs assertion. Verified by executing the notebook end to end (training job Completed).
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.