Add files using upload-large-folder tool

.gitattributes

@@ -58,3 +58,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+metadata/train.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,194 @@
+---
+pretty_name: AV-Dense-60k
+language:
+- en
+task_categories:
+- text-to-video
+- text-to-audio
+- video-text-to-text
+tags:
+- audio-video
+- multimodal
+- dense-captioning
+- reinforcement-learning
+- webdataset
+license: other
+---
+
+# AV-Dense-60k
+
+AV-Dense-60k contains paired video, audio, raw text, and dense audio-video captions.
+The media is packaged as deterministic WebDataset TAR shards, with portable metadata
+and checksums for streaming or restoration.
+
+> **Rights and license notice:** this repository uses `license: other`. Public
+> availability does not relicense the underlying media and does not grant rights that
+> the publisher does not hold. Before downloading, redistributing, or using any sample,
+> users must independently confirm the media's provenance, copyright and neighboring
+> rights, platform terms, privacy/consent requirements, and suitability for their
+> intended jurisdiction and use.
+
+## Verified dataset inventory
+
+| Split | Records | Video | Audio | Intended role |
+| --- | ---: | ---: | ---: | --- |
+| train | 44,178 | 44,178 | 44,178 | model training |
+| validation | 5,522 | 5,522 | 5,522 | held-out validation |
+| test | 5,523 | 5,523 | 5,523 | final evaluation only |
+
+The split IDs are unique and pairwise disjoint. Every manifest ID has exactly one MP4
+and one WAV in `processed_gt_data`; no missing, orphan, or zero-byte files were found.
+`AV60KDenseBench.csv` contains the same 5,523 IDs as `test.jsonl` and must not be used
+for training. The additional `480p_grpo` variant contains the 5,522 validation IDs.
+
+The upstream metadata contained obsolete machine-specific media locations. Release
+staging reconstructs every video/audio field from the split and `video_id`/`id`, so all
+structured paths in this repository are canonical and relative to the restored dataset
+root, for example `processed_gt_data/train/videos/<video_id>.mp4`.
+
+## Repository contents
+
+This data repository contains only the following release surfaces:
+
+```text
+README.md
+metadata/
+  train.jsonl
+  val.jsonl
+  test.jsonl
+  AV60KDenseBench.csv
+  AV69kDense-vanilla-max4s.pt
+  align_csv.py
+data/
+  processed/train/avdense-train-*.tar
+  processed/val/avdense-val-*.tar
+  processed/test/avdense-test-*.tar
+  legacy_480p_grpo/avdense-val480p-*.tar
+manifests/
+  shards.json
+  SHA256SUMS
+```
+
+Training, extraction, and manifest-building code is distributed separately from this
+dataset repository. No model source, checkpoint, experiment report, or presentation is
+claimed as part of this public data release.
+
+## WebDataset sample format
+
+Each processed sample has three same-stem members:
+
+```text
+<video_id>.mp4
+<video_id>.wav
+<video_id>.json
+```
+
+The JSON member contains `sample_id`, `split`, `variant`, `prompt_av`, `prompt_v`,
+`prompt_a`, `video_path`, and `audio_path`. `prompt_av`/`prompt_v` use
+`dense_caption`; `prompt_a` uses `raw_text`; both media paths are relative.
+
+## Stream shards without extraction
+
+Install the optional `webdataset` package, then run this example from the downloaded
+data repository root:
+
+```bash
+python -m pip install webdataset
+python - <<'PY'
+import json
+from pathlib import Path
+
+import webdataset as wds
+
+shards = sorted(Path("data/processed/train").glob("avdense-train-*.tar"))
+if not shards:
+    raise SystemExit("no train shards found; run from the data repository root")
+
+samples = wds.WebDataset([str(path) for path in shards], shardshuffle=False).to_tuple(
+    "__key__", "mp4", "wav", "json"
+)
+for sample_id, mp4_bytes, wav_bytes, metadata_bytes in samples:
+    metadata = json.loads(metadata_bytes)
+    print(sample_id, len(mp4_bytes), len(wav_bytes), metadata["prompt_av"])
+    break
+PY
+```
+
+## Integrity verification
+
+Before extraction, run the checksum gate from the data repository root:
+
+```bash
+sha256sum --check manifests/SHA256SUMS
+```
+
+Do not extract or train if any shard is missing or is not reported as `OK`.
+
+For a stronger, read-only audit, use `extract_avdense_webdataset.py` from the separately
+distributed JavisR1 code package. The `--verify-only` mode rechecks shard digests,
+manifest totals, TAR structure, sample counts, relative JSON paths, and every
+MP4/WAV/JSON triplet without writing output.
+
+One reproducible relative workspace arrangement is:
+
+```text
+workspace/
+  AV-Dense-60k/          # this data repository
+  code/jarvisr1/         # separately distributed code package
+  datasets/              # restoration destination
+```
+
+From `workspace/code/jarvisr1`:
+
+```bash
+python3 scripts/extract_avdense_webdataset.py \
+  --repo-root ../../AV-Dense-60k \
+  --verify-only
+```
+
+## Restore the training layout
+
+Run restoration only after both checksum and verify-only gates succeed. From the same
+separately distributed code-package directory:
+
+```bash
+python3 scripts/extract_avdense_webdataset.py \
+  --repo-root ../../AV-Dense-60k \
+  --output-root ../../datasets/AV-Dense-60k
+
+python3 scripts/build_avdense_manifests.py \
+  --dataset-root ../../datasets/AV-Dense-60k
+```
+
+The extractor restores the normalized metadata plus the following default media tree:
+
+```text
+../../datasets/AV-Dense-60k/
+  train.jsonl
+  val.jsonl
+  test.jsonl
+  processed_gt_data/train/{videos,audios}/...
+  processed_gt_data/val/{videos,audios}/...
+  processed_gt_data/test/{videos,audios}/...
+```
+
+To additionally restore the legacy validation-resolution variant, add
+`--include-legacy-480p` to the extraction command. The canonical builder produces:
+
+```text
+sample_id,prompt_av,prompt_v,prompt_a,video_path,audio_path,split
+```
+
+## Responsible use and limitations
+
+- `license: other` is intentional; do not infer a permissive media license from public
+  repository visibility.
+- Users are responsible for rights clearance, attribution obligations, privacy review,
+  consent, and compliance with applicable law and source-platform terms.
+- Keep train, validation, and test IDs disjoint. Test and benchmark rows are evaluation
+  only.
+- Dense captions may contain factual, temporal, or synchronization errors and should
+  not be treated as ground truth without task-appropriate review.
+- Dataset inclusion does not imply endorsement of a depicted person, action, statement,
+  or downstream application.
+- This repository makes no model-quality claim.

data/legacy_480p_grpo/avdense-val480p-00000.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6feaa82b8c187413641c6b8e3b7bb262e90f13c87c3b92e9a28d1f2595e84cdc
+size 1077084160

data/legacy_480p_grpo/avdense-val480p-00001.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c00f434f5ae510a8a366369a6cdf16e48ac1558cb566716da872e33984b10dca
+size 1076981760

data/legacy_480p_grpo/avdense-val480p-00002.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71cd2ad99198408de371da2894d1de1e30304819b6a71158201cfd5d2135b1b9
+size 1077053440

data/legacy_480p_grpo/avdense-val480p-00003.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f426672a48f84c9120d97fd8705643c41fd9e66e8a3d0239ed34ae94f3a87156
+size 633774080

data/processed/test/avdense-test-00000.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abf971bf075466e9fbea87ab536e9cd91261c2aceb469c3b57bd0a704de02906
+size 1084876800

data/processed/test/avdense-test-00001.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec59bb4067bb58e59f98661b4b74a4262b517f23dcf5262d703e8dde5aa8f9b3
+size 235571200

data/processed/train/avdense-train-00000.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c811ee2fec92cd985241d4f69fe58ae3294a00d29ce5e36027860e1a2d7f923
+size 1084856320

data/processed/train/avdense-train-00001.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00e6714cd779d46ee688fe0af03a0d56a3c3bb31933d899b2ec99d9fcf9103e5
+size 1084887040

data/processed/train/avdense-train-00002.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36151abddf837994ce8fb1000a671254ed9c9a0a2eafcafb862ccfe28590581e
+size 1084856320

data/processed/train/avdense-train-00003.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e430946db1d6a561ec677f164261f5347a22cbbfbc1ed82b85aac4d35865a5b
+size 1084999680

data/processed/train/avdense-train-00004.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f7c9ecaec976e51d0455aa557716979ce5182a943df71ddc906de951a0689c0
+size 1084917760

data/processed/train/avdense-train-00005.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f68132b921ccab345647e87e73a14977d1e6a0e107d22881a59926aa0af4c41
+size 1084928000

data/processed/train/avdense-train-00006.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b66b91f03e25f6140908c35eef06a1de9710f712e7a40cd329d7c1f6af317031
+size 1084723200

data/processed/train/avdense-train-00007.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1abea16b9135c793e2c91a950ad3b12d3726d04b368f0e97a6774e8c3008fd51
+size 1084815360

data/processed/train/avdense-train-00008.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34c3c07a5391c5823eb8cb0724c71ab7a3f01305ff36cea1d59b7d502cd25755
+size 1084835840

data/processed/train/avdense-train-00009.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86591e76d006a3c5e967240aca7ea63b9abbde6d005398353cd2370bcde2bf9a
+size 847697920

data/processed/val/avdense-val-00000.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4332506fcebb7eea48e02f6e2fd88306302d05e13541dc817a13caee69d05cb6
+size 1084948480

data/processed/val/avdense-val-00001.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe945cceee14109f36cc9b56c936677bb39e1d0cc22d9a8fbdb1e28e21abba53
+size 237137920

manifests/SHA256SUMS

@@ -0,0 +1,18 @@
+2c811ee2fec92cd985241d4f69fe58ae3294a00d29ce5e36027860e1a2d7f923  data/processed/train/avdense-train-00000.tar
+00e6714cd779d46ee688fe0af03a0d56a3c3bb31933d899b2ec99d9fcf9103e5  data/processed/train/avdense-train-00001.tar
+36151abddf837994ce8fb1000a671254ed9c9a0a2eafcafb862ccfe28590581e  data/processed/train/avdense-train-00002.tar
+3e430946db1d6a561ec677f164261f5347a22cbbfbc1ed82b85aac4d35865a5b  data/processed/train/avdense-train-00003.tar
+4f7c9ecaec976e51d0455aa557716979ce5182a943df71ddc906de951a0689c0  data/processed/train/avdense-train-00004.tar
+5f68132b921ccab345647e87e73a14977d1e6a0e107d22881a59926aa0af4c41  data/processed/train/avdense-train-00005.tar
+b66b91f03e25f6140908c35eef06a1de9710f712e7a40cd329d7c1f6af317031  data/processed/train/avdense-train-00006.tar
+1abea16b9135c793e2c91a950ad3b12d3726d04b368f0e97a6774e8c3008fd51  data/processed/train/avdense-train-00007.tar
+34c3c07a5391c5823eb8cb0724c71ab7a3f01305ff36cea1d59b7d502cd25755  data/processed/train/avdense-train-00008.tar
+86591e76d006a3c5e967240aca7ea63b9abbde6d005398353cd2370bcde2bf9a  data/processed/train/avdense-train-00009.tar
+4332506fcebb7eea48e02f6e2fd88306302d05e13541dc817a13caee69d05cb6  data/processed/val/avdense-val-00000.tar
+fe945cceee14109f36cc9b56c936677bb39e1d0cc22d9a8fbdb1e28e21abba53  data/processed/val/avdense-val-00001.tar
+abf971bf075466e9fbea87ab536e9cd91261c2aceb469c3b57bd0a704de02906  data/processed/test/avdense-test-00000.tar
+ec59bb4067bb58e59f98661b4b74a4262b517f23dcf5262d703e8dde5aa8f9b3  data/processed/test/avdense-test-00001.tar
+6feaa82b8c187413641c6b8e3b7bb262e90f13c87c3b92e9a28d1f2595e84cdc  data/legacy_480p_grpo/avdense-val480p-00000.tar
+c00f434f5ae510a8a366369a6cdf16e48ac1558cb566716da872e33984b10dca  data/legacy_480p_grpo/avdense-val480p-00001.tar
+71cd2ad99198408de371da2894d1de1e30304819b6a71158201cfd5d2135b1b9  data/legacy_480p_grpo/avdense-val480p-00002.tar
+f426672a48f84c9120d97fd8705643c41fd9e66e8a3d0239ed34ae94f3a87156  data/legacy_480p_grpo/avdense-val480p-00003.tar

manifests/shards.json

@@ -0,0 +1,170 @@
+{
+  "format": "avdense-webdataset-v1",
+  "shard_target_bytes": 1073741824,
+  "shards": [
+    {
+      "path": "data/processed/train/avdense-train-00000.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4506,
+      "source_bytes": 1073666998,
+      "tar_bytes": 1084856320,
+      "sha256": "2c811ee2fec92cd985241d4f69fe58ae3294a00d29ce5e36027860e1a2d7f923"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00001.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4494,
+      "source_bytes": 1073719749,
+      "tar_bytes": 1084887040,
+      "sha256": "00e6714cd779d46ee688fe0af03a0d56a3c3bb31933d899b2ec99d9fcf9103e5"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00002.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4499,
+      "source_bytes": 1073687499,
+      "tar_bytes": 1084856320,
+      "sha256": "36151abddf837994ce8fb1000a671254ed9c9a0a2eafcafb862ccfe28590581e"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00003.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4536,
+      "source_bytes": 1073739519,
+      "tar_bytes": 1084999680,
+      "sha256": "3e430946db1d6a561ec677f164261f5347a22cbbfbc1ed82b85aac4d35865a5b"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00004.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4504,
+      "source_bytes": 1073716748,
+      "tar_bytes": 1084917760,
+      "sha256": "4f7c9ecaec976e51d0455aa557716979ce5182a943df71ddc906de951a0689c0"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00005.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4524,
+      "source_bytes": 1073692400,
+      "tar_bytes": 1084928000,
+      "sha256": "5f68132b921ccab345647e87e73a14977d1e6a0e107d22881a59926aa0af4c41"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00006.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4529,
+      "source_bytes": 1073523657,
+      "tar_bytes": 1084723200,
+      "sha256": "b66b91f03e25f6140908c35eef06a1de9710f712e7a40cd329d7c1f6af317031"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00007.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4516,
+      "source_bytes": 1073579454,
+      "tar_bytes": 1084815360,
+      "sha256": "1abea16b9135c793e2c91a950ad3b12d3726d04b368f0e97a6774e8c3008fd51"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00008.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 4517,
+      "source_bytes": 1073603660,
+      "tar_bytes": 1084835840,
+      "sha256": "34c3c07a5391c5823eb8cb0724c71ab7a3f01305ff36cea1d59b7d502cd25755"
+    },
+    {
+      "path": "data/processed/train/avdense-train-00009.tar",
+      "split": "train",
+      "variant": "processed_gt_data",
+      "samples": 3553,
+      "source_bytes": 838855194,
+      "tar_bytes": 847697920,
+      "sha256": "86591e76d006a3c5e967240aca7ea63b9abbde6d005398353cd2370bcde2bf9a"
+    },
+    {
+      "path": "data/processed/val/avdense-val-00000.tar",
+      "split": "val",
+      "variant": "processed_gt_data",
+      "samples": 4535,
+      "source_bytes": 1073687191,
+      "tar_bytes": 1084948480,
+      "sha256": "4332506fcebb7eea48e02f6e2fd88306302d05e13541dc817a13caee69d05cb6"
+    },
+    {
+      "path": "data/processed/val/avdense-val-00001.tar",
+      "split": "val",
+      "variant": "processed_gt_data",
+      "samples": 987,
+      "source_bytes": 234670729,
+      "tar_bytes": 237137920,
+      "sha256": "fe945cceee14109f36cc9b56c936677bb39e1d0cc22d9a8fbdb1e28e21abba53"
+    },
+    {
+      "path": "data/processed/test/avdense-test-00000.tar",
+      "split": "test",
+      "variant": "processed_gt_data",
+      "samples": 4543,
+      "source_bytes": 1073572721,
+      "tar_bytes": 1084876800,
+      "sha256": "abf971bf075466e9fbea87ab536e9cd91261c2aceb469c3b57bd0a704de02906"
+    },
+    {
+      "path": "data/processed/test/avdense-test-00001.tar",
+      "split": "test",
+      "variant": "processed_gt_data",
+      "samples": 980,
+      "source_bytes": 233126473,
+      "tar_bytes": 235571200,
+      "sha256": "ec59bb4067bb58e59f98661b4b74a4262b517f23dcf5262d703e8dde5aa8f9b3"
+    },
+    {
+      "path": "data/legacy_480p_grpo/avdense-val480p-00000.tar",
+      "split": "val",
+      "variant": "480p_grpo",
+      "samples": 1503,
+      "source_bytes": 1073503562,
+      "tar_bytes": 1077084160,
+      "sha256": "6feaa82b8c187413641c6b8e3b7bb262e90f13c87c3b92e9a28d1f2595e84cdc"
+    },
+    {
+      "path": "data/legacy_480p_grpo/avdense-val480p-00001.tar",
+      "split": "val",
+      "variant": "480p_grpo",
+      "samples": 1550,
+      "source_bytes": 1073289772,
+      "tar_bytes": 1076981760,
+      "sha256": "c00f434f5ae510a8a366369a6cdf16e48ac1558cb566716da872e33984b10dca"
+    },
+    {
+      "path": "data/legacy_480p_grpo/avdense-val480p-00002.tar",
+      "split": "val",
+      "variant": "480p_grpo",
+      "samples": 1572,
+      "source_bytes": 1073316912,
+      "tar_bytes": 1077053440,
+      "sha256": "71cd2ad99198408de371da2894d1de1e30304819b6a71158201cfd5d2135b1b9"
+    },
+    {
+      "path": "data/legacy_480p_grpo/avdense-val480p-00003.tar",
+      "split": "val",
+      "variant": "480p_grpo",
+      "samples": 897,
+      "source_bytes": 631632014,
+      "tar_bytes": 633774080,
+      "sha256": "f426672a48f84c9120d97fd8705643c41fd9e66e8a3d0239ed34ae94f3a87156"
+    }
+  ],
+  "total_samples_including_legacy_variant": 60745,
+  "total_tar_bytes": 17118945280
+}

metadata/AV69kDense-vanilla-max4s.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8f1077b526446e2e04a77ec675404310c3e2b16ac68a95ad1e07573170d2d41
+size 31460552

metadata/align_csv.py

@@ -0,0 +1,71 @@
+import pandas as pd
+import os
+import sys
+
+def align_csv_structure(csv1_path, csv2_path):
+    """
+    Reads csv1 and csv2, reformats csv1 to match csv2's columns, 
+    and overwrites csv1.
+    """
+    
+    # 1. 读取 CSV 文件
+    try:
+        df1 = pd.read_csv(csv1_path)
+        df2 = pd.read_csv(csv2_path)
+    except FileNotFoundError as e:
+        print(f"Error: {e}")
+        return
+
+    # 获取目标列结构 (来自 csv2)
+    target_columns = df2.columns.tolist()
+    
+    # 2. 遍历目标列，处理 csv1 中缺失的列
+    for col in target_columns:
+        if col in df1.columns:
+            # 如果 csv1 已经有这个列，跳过（保留原数据）
+            continue
+        
+        # 如果 csv1 缺少这个列，根据 pattern 进行生成或填充默认值
+        if col == 'id':
+            # Pattern: 从 path 中提取文件名（不带扩展名）
+            df1['id'] = df1['path'].apply(lambda x: os.path.splitext(os.path.basename(x))[0] if pd.notnull(x) else '')
+            
+        elif col == 'relpath':
+            # Pattern: 从 path 中提取完整文件名
+            df1['relpath'] = df1['path'].apply(lambda x: os.path.basename(x) if pd.notnull(x) else '')
+            
+        elif col == 'audio_id':
+            # Pattern: 从 audio_path 中提取文件名（不带扩展名）
+            if 'audio_path' in df1.columns:
+                df1['audio_id'] = df1['audio_path'].apply(lambda x: os.path.splitext(os.path.basename(x))[0] if pd.notnull(x) else '')
+            else:
+                df1['audio_id'] = ''
+                
+        elif col == 'audio_fps':
+            # 默认填充 16000 (参考 csv2 的常见值，或者是 NaN)
+            df1['audio_fps'] = 16000
+            
+        # 处理 Category Title 列 (例如 cat0_title)
+        elif 'title' in col and col.startswith('cat'):
+            # 由于没有 ID 到 Title 的映射表，这里留空
+            df1[col] = ''
+            
+        # 其他技术参数 (num_frames, height, width, fps, aes, flow, ocr 等)
+        else:
+            # 填充为空值 (NaN)，表示数据缺失
+            df1[col] = pd.NA
+
+    # 3. 重新排序并筛选列，确保与 csv2 完全一致
+    # reindex 会自动丢弃 csv1 中多余的列（如果 csv2 没有的话），并对齐顺序
+    df1_final = df1.reindex(columns=target_columns)
+
+    # 4. 覆盖保存 csv1
+    df1_final.to_csv(csv1_path, index=False)
+    print(f"Successfully processed and overwritten: {csv1_path}")
+    print(f"Columns aligned: {len(target_columns)}")
+
+if __name__ == "__main__":
+    if len(sys.argv) != 3:
+        print("Usage: python align_csv.py <csv-to-align> <reference-schema-csv>", file=sys.stderr)
+        raise SystemExit(2)
+    align_csv_structure(sys.argv[1], sys.argv[2])

metadata/train.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8e95edbf670be2492c76da56d7babd9364122828bee0a2e9b83f2553888d583
+size 52614917