Fix model-weight distribution so the repo is clone-and-run#1
Open
mb-westover wants to merge 1 commit into
Open
Fix model-weight distribution so the repo is clone-and-run#1mb-westover wants to merge 1 commit into
mb-westover wants to merge 1 commit into
Conversation
The weight caches (pd_channel_cache, cet_cache, dl_cache, models) are gitignored, but the README claimed they ship in git — so a fresh clone could not run the models, and the .gitignore pointed at an empty mirror path (iiic-freq3/data). Upload these ~12 MB inference weights to s3://bdsp-opendata-credentialed/grond/data/ (verified round-trip), document the sync step in the README, and correct the .gitignore mirror path. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
A fresh
git cloneof GROND could not run the models. The weight caches(
pd_channel_cache,cet_cache,dl_cache,models) are gitignored, but theREADME claimed they ship in git, and
.gitignorepointed at an empty mirrorpath (
s3://bdsp-opendata-credentialed/iiic-freq3/data/). So the ~12 MB ofinference weights a clone needs were unreachable from any documented source.
Fix
pd_channel_cache,cet_cache,dl_cache/ridge_freq_model.npz,models/bipd_gbt.pkl) tos3://bdsp-opendata-credentialed/grond/data/. Round-trip verified: a freshaws s3 syncfetches all 5cnn_attnfolds + the ridge model and they load.sync step (
aws s3 sync s3://…/grond/data/ data/) and updated the S3 contents..gitignore: corrected the staleiiic-freq3/datamirror path togrond/data/and documented the sync command.Verified: all 11 models (UnifiedPDModel, HemiCET, ChannelPD-Net, CET-UNet,
PDProfiler, RDA-Profiler, ridge, BIPD-GBT, …) load and run after
git clone+aws s3 sync s3://…/grond/data/ data/.Notes (not addressed here)
e2efolds 3–4 exist only as mid-training checkpoints (experimental, not inthe published pipeline).
code/visualization/visualize_embeddings.pyis orphaned (imports a nonexistentcode/dl/module) — candidate for removal.🤖 Generated with Claude Code