Fix cached latents configs#336
Merged
sgreenbury merged 3 commits intomainfrom Apr 20, 2026
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This pull request introduces robust preflight validation to ensure cached latent-space data is always consistent with the corresponding autoencoder (AE) training configuration. It adds explicit normalization settings to all cache-latents experiment configs, and updates all relevant SLURM submission scripts to fail fast if there is any mismatch in critical datamodule fields between the cache and AE configs. The main changes are grouped below:
Validation and Consistency Enforcement:
validate_cached_latents_against_ae.sh, which checks that critical datamodule fields (such asdata_path,n_steps_input,stride,use_normalization, etc.) in cached-latents configs match those in the AE training config, and that normalization usage is explicitly set and consistent. This script is now sourced and used by all cached-latents submission scripts.submit_cache_latents.sh,submit_crps_latent_large.sh,submit_crps_latent_timing.sh,submit_crps_latent_cns_large.sh,submit_crps_latent_cns_timing.sh,submit_fm_large.sh,submit_fm_timing.sh) now invoke this validation and will skip or abort if configs are inconsistent. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]Experiment Configuration Updates:
datamodule.use_normalization: truein all cache-latents experiment YAMLs, ensuring that normalization intent is always clear and validated. [1] [2] [3] [4]Documentation:
slurm_scripts/comparison/README.mdto document the new validation checks and clarify that the cache-latents experiment configs are the source of truth for normalization and other datamodule settings. [1] [2]These changes collectively prevent silent mismatches between cached latents and AE configs, improving experiment reproducibility and reliability.