Generate documents with Zephyr#19
Open
alxmrs wants to merge 19 commits into
Open
Conversation
alxmrs
force-pushed
the
u/alxmrs/zyph-v1-contacts
branch
from
db370ec to
16efdb7
Compare
alxmrs
commented
May 27, 2026
| [tool.uv.sources] | ||
| marinfold = { path = "../../marinfold", editable = true } | ||
| marinfold = { git = "https://github.com/Open-Athena/MarinFold.git", subdirectory = "marinfold" } | ||
| marin-zephyr = { git = "https://github.com/marin-community/marin.git", subdirectory = "lib/zephyr", branch = "alxmrs/stamp-iris-build-date" } |
Member
Author
There was a problem hiding this comment.
This branch is necessary to not hit a Iris client issue.
Member
Author
|
@codex may I have your review? |
Make --num-docs limit the number of input files processed (one doc per
structure, ≈ N docs) rather than post-filtering emitted docs. This
restores the original generate_documents semantics and bounds parsing to
the first N files. Sharding is driven by the --out pattern: a {shard}
placeholder writes one parquet per input, otherwise a single merged file.
- parse.list_structure_files(): brace-expanding glob resolver with a limit
- cli: from_iterable(first N files) when --num-docs is set, else from_files
- README: corrected single-line Iris command + arg/prefix/cancel guidance
- tests: --num-docs now counts inputs (unparseable inputs consume budget)
and spreads one-per-file under a {shard} --out
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add a second input modality so --input can point at the timodonnell/afdb-1.6M dataset (parquet shards whose cif_content column holds raw mmCIF text), in addition to .cif/.pdb files. Dispatched on the --input extension: - .parquet -> from_files(glob).load_parquet([cif_column, entry_id]) and parse the text in memory via parse.parse_cif_content. entry_id is loaded because generate_one seeds its RNG from it. - otherwise -> the existing file-path + gemmi-read path. parse.parse_structure is refactored to share _build_parsed_structure with the new parse_cif_content / try_parse_cif_content. Adds a --cif-column flag (default cif_content) and huggingface_hub as a base dep for hf://. README documents the hf:// smoke-test (single shard) and full-run globs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
alxmrs
force-pushed
the
u/alxmrs/zyph-v1-contacts
branch
from
176baca to
a8c014d
Compare
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.
Update exp1 to generate documents on Zephyr and Iris at scale. Fixes #5.
We can verify the output document quality here:
gs://marin-us-east5/protein-structure/MarinFold/exp1/corpus_v1-{shard:05d}-of-{total:05d}.parquet.