[agents doc] notes on when to create new blocksets for checkpoint variant#14208
Open
yiyixuxu wants to merge 6 commits into
Open
[agents doc] notes on when to create new blocksets for checkpoint variant#14208yiyixuxu wants to merge 6 commits into
yiyixuxu wants to merge 6 commits into
Conversation
Contributors keep handling checkpoint variants (distilled/turbo) with a ConfigSpec flag and an if-branch inside a shared block — the standard-pipeline mindset ported into modular. The doc never said what to do instead: nothing routed on 'behavior differs per checkpoint', and flux2-klein was only cited as a flatness example, not as the variant pattern it embodies. Adds: - a per-checkpoint branch in the block types decision tree - 'Key pattern: Checkpoint variants' — separate assembly when the variant changes the contract (inputs/components/blocks), ConfigSpec only when it changes a value; flux2-klein as reference; the config-flag anti-pattern - gotcha #9 for the if-components.config.<variant_flag> smell Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…pos) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ponents/config, never inputs) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The config-flag branch is what standard pipelines are forced into; modular's pitch is that the loaded pipeline describes exactly its checkpoint. Lead with the payoff (clean per-variant contract, automatic routing, input-surface expressiveness) and present the flag branch as tolerable-but-lossy rather than forbidden. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ckset unless literally the same Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…alize only by removing branches/duplicates Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
sayakpaul
approved these changes
Jul 17, 2026
sayakpaul
left a comment
sayakpaul
left a comment
Member
There was a problem hiding this comment.
Thanks. Nothing major from my end.
| NO -> ConditionalPipelineBlocks (custom select_block method) | ||
|
|
||
| Is it a different CHECKPOINT (distilled / turbo / a variant with its own schedule)? | ||
| YES -> its own blocks assembly, unless it behaves literally the same |
Member
There was a problem hiding this comment.
"its own block assembly" reads a bit confusing to me. Possible reword?
| 3. `before_denoise.py` -- Timesteps, latent prep, noise setup. Each logical operation = one block | ||
| 4. `denoise.py` -- The hardest. Convert guidance to guider abstraction | ||
|
|
||
| ## Growing a pipeline: one workflow at a time |
Member
There was a problem hiding this comment.
Possible to also link an example here?
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.
Keeping seeing modular PRs not creating separate blocksets for distilled checkpoints, so adding notes on this
#14083
#14177