bghira's picture
Add complete pinned ScaleEdit-12M metadata companion for native Parquet loading
62cf59a verified
|
Raw History Blame Contribute Delete
4.23 kB
metadata
license: cc-by-nc-sa-4.0
task_categories:
  - image-to-image
language:
  - en
size_categories:
  - 10M<n<100M
tags:
  - webshart
  - parquet
  - metadata-only
pretty_name: ScaleEdit-12M Webshart Metadata
configs:
  - config_name: default
    data_files:
      - split: train
        path: metadata/*/*.parquet

ScaleEdit-12M: metadata-only Webshart companion

This repository contains 11,845,023 metadata records across 251 Parquet shards covering every row of InternVL-U/ScaleEdit-12M, pinned to commit 05d2eea7b80ba72c077d39be43b7aca1a7857e6a. It contains no image payloads and does not duplicate the source's 6,188,768,447,762 bytes of compressed Parquet. The metadata Parquet files total 423,499,152 bytes.

Webshart usage

Requires the Webshart source build that exposes ParquetDataset and ParquetDataLoader. The published webshart 0.6.0 wheel does not support this format. Install that source checkout with pip install -e '.[hub]'; until the Parquet feature is released, use PyArrow to read these metadata files independently. This is not a tar byte-offset index.

import webshart

dataset = webshart.discover_dataset(
    "InternVL-U/ScaleEdit-12M",
    format="parquet",
    metadata="webshart/ScaleEdit-12M-metadata",
)
loader = webshart.ParquetDataLoader(dataset, image_columns=["source_image", "edited_image"])
metadata = loader.get_metadata(0, 0)
sample = loader.load_sample(0, 0)
edited_bytes = sample.data
source_bytes = sample.payloads["source_image"]
instruction = sample.captions

Prefer loader.iter_shard(shard_index) for sequential training. The source has very large compressed row groups and dictionaries: native page streaming avoids loading an entire 15+ GB image column, but initial dictionary decoding can still cost hundreds of MiB. Random row access scans preceding rows within its row group; this is not constant-cost tar-style seeking. Pass only the image columns you need to limit memory and network use.

webshart-parquet.json maps source shards to metadata files, pins the upstream revision, records row counts/file sizes, and verifies metadata SHA-256 checksums. metadata_mode="replace" uses these complete companion rows without reading upstream metadata columns. The manifest selects edited-image dimensions and maps edit_instruction to captions; it does not invent richer JSON captions. Custom companion files can store native struct/list captions or declare json_columns for JSON-encoded mixed string/object caption lists.

Schema and identity

Every original non-image column is preserved: id, edit_task, edit_instruction, source URL/SHA-256/fetch-date fields, source and edited width/height, and all three quality scores. Added locators are source_path, source_row (zero-based absolute row within the source file), and source_row_group. IDs repeat between task categories; join by pinned source path plus row position, not id alone.

Source images for 702,078 rows are referenced by URLs instead of embedded bytes. Those URLs, hashes, and fetch dates are preserved, but no external image URL was fetched or rehosted. Webshart returns None for absent source bytes; applications must explicitly implement their own retrieval/verification policy. As upstream explains, original linked-image rights remain with their owners.

Provenance and validation

Generated on a CPU-only pod using projected non-image reads, not a full dataset download. Projection read 508,945,853 logical range bytes. Each output was compared against its projected Arrow table and SHA-256 checked; all source paths, row positions, dimensions, counts, and shard coverage were validated. summary.json contains per-task counts and totals. Metadata uses Zstd compression, 2,048-row groups, and page indexes. No source or edited image bytes are included in this repository.

See the upstream dataset card for dataset construction, limitations, attribution, and citation. This metadata derivative retains the upstream CC-BY-NC-SA-4.0 license and is not an endorsement by the original authors. Upstream usage restrictions still apply.