Add portable Level3 instructions, video references and GEM input reconstruction

level3/README.md

@@ -0,0 +1,94 @@
+# RoboSteer Level3 preprocessing assets
+
+This folder releases three Level3 assets as portable tables and a reconstruction tool. It is independent of the sibling `level1/` assets. The original server paths and video symlinks are not distributed.
+
+## Assets
+
+| Folder | Contents | Reconstructed layout |
+|---|---|---|
+| `text_audio_instructions/` | 17,575 UTF-8 instructions across 7,764 tasks, including original audio/text modality and order | `<task_id>/<item_id>__text.txt` or `__audio.txt` |
+| `video_symlinks/` | 8,475 video references with paths relative to a user-provided public-data root, byte sizes and SHA-256 | `<item_id>.mp4` |
+| `interleaved_input_list_for_GEM/` | 26,050 ordered items: 17,575 text items and 8,475 video references | `<task_id>/order_NNN.txt` or `.mp4` |
+
+No raw audio transcripts, audio files, image assets or video bytes are included. The audio-origin instructions are processed motion instructions, not raw Whisper transcripts. The original GEM preparation skipped 7,764 image items. Original order numbers are preserved, including gaps left by skipped images.
+
+`release_manifest.json` records the comparison between the grouped text and the actual GEM text. If differences exist, `text_overrides.parquet` preserves them for exact reconstruction; the grouped instruction table remains unchanged.
+
+## Download
+
+```bash
+pip install huggingface_hub pyarrow
+hf download PhoebeCC/RoboSteer-Preprocessing \
+  --repo-type dataset --include 'level3/*' --local-dir ./RoboSteer-Preprocessing
+```
+
+For reproducibility, add `--revision <commit-id>` using the commit of the Level3 release.
+
+## External video dependency
+
+Prepare the matching RoboSteer public data locally:
+
+```text
+<public-root>/
+└── Data_Processed/Level3/Video_Slices/
+    └── <original segment filename>.mp4
+```
+
+The release identifies every required video using `source_relpath`, `source_size` and `source_sha256`. A matching folder layout alone is insufficient: each video must also match its recorded bytes. This release does not download or regenerate the original media.
+
+## Reconstruct the three original assets
+
+```bash
+python RoboSteer-Preprocessing/level3/scripts/materialize_level3.py \
+  --public-root /data/public \
+  --output-root ./restored_level3 \
+  --video-mode symlink
+```
+
+The result is:
+
+```text
+restored_level3/
+├── text_audio_instructions/
+├── video_symlinks/
+└── interleaved_input_list_for_GEM/
+```
+
+This reproduces the selected assets directly under the output root. To match the organized project layout, pass `--output-root /path/to/cxt/level3/inputs` instead. Use a new output directory for initial validation.
+
+The tool verifies release checksums and all source videos before writing. It writes exact UTF-8 text bytes, including existing newlines. It does not overwrite conflicting files. Existing matching files are accepted. Missing or different source videos produce a JSON error report on stdout and exit code 2. Conflicting or missing reconstructed outputs in verification mode produce exit code 3. Original input videos remain untouched.
+
+Use `--video-mode copy` if symlinks are unavailable. This copies video bytes into both video layouts and therefore uses additional disk space. Default symlinks point to the supplied local public-data root; moving that root later requires rebuilding them.
+
+Verify the restored assets:
+
+```bash
+python RoboSteer-Preprocessing/level3/scripts/materialize_level3.py \
+  --public-root /data/public \
+  --output-root ./restored_level3 \
+  --verify-only
+```
+
+Verification checks all expected entries; it does not remove or reject unrelated extra files. It does not run GEM or evaluate video decoding or instruction quality.
+
+## Read the tables directly
+
+```python
+import pyarrow.parquet as pq
+
+instructions = pq.read_table(
+    "RoboSteer-Preprocessing/level3/text_audio_instructions/data.parquet"
+).to_pylist()
+```
+
+- Instructions: `item_id`, `task_id`, `order`, `source_modality`, `file_name`, `text`, `sha256`.
+- Videos: `item_id`, `task_id`, `order`, `file_name`, `source_relpath`, `source_size`, `source_sha256`.
+- GEM items: `task_id`, `item_id`, `order`, `file_name`, `kind`, `sha256`, `source_relpath` (null for text). Text is joined by `item_id`; optional overrides take precedence only for GEM reconstruction.
+
+`file_name` is relative to its reconstructed asset directory. Media paths are relative to `--public-root`, never to the HF cache. The video table contains references, not embedded `Video` objects; viewing it on Hugging Face does not make external videos playable.
+
+## Validation and provenance
+
+The text was exported from the existing Level3 preprocessing outputs. Video references were taken from the original symlink targets, and the task/order layout from the actual GEM input directories. Text and video integrity are recorded as SHA-256; no historical processing `status` values were reused as quality labels.
+
+See `release_manifest.json`, `validation.json` and `SHA256SUMS`. The reconstruction utility requires Python 3.9+ and pyarrow. The release does not include Qwen/Whisper/GEM models, logs, server credentials, or raw absolute server paths. Dataset terms and upstream media conditions should be consulted at the repository and source-data level.

level3/SHA256SUMS

@@ -0,0 +1,10 @@
+e714770f36f1e3e961d249ab8d444ec974662ad14a23afb8158f658bc958e637  README.md
+b2b23339bf9e2046f36c38e71442d14964cbd7ce812458b95c565547a697574c  interleaved_input_list_for_GEM/README.md
+8f4222a67e2f2cdefbdadb3f9ed18dd296835e969e1823a693ed48edea6c2eb9  interleaved_input_list_for_GEM/items.parquet
+25847789e2be0271d91f247dc161a765259d421e04734dba8b65f408f133e826  release_manifest.json
+6c302570f29b760392f9b259e02730ecfbe9b6001d64671ce211197c50e79e11  scripts/materialize_level3.py
+c0f1fcbc3362a74d11afefb0508168fd729150023c101bd4221de26afb331f1f  text_audio_instructions/README.md
+1771747ae54f8c1e19e9973e5971bfe3f89293d63e6983d3a320ccdb562e7605  text_audio_instructions/data.parquet
+7816e8cd76f178e9b3b37bca47f3a7931c37610e8c53316539a9f2720063907a  validation.json
+0d6c9f78ba8b73604e7d1c979f16b67069bd15d165308d0b58d432a17c6b3e2a  video_symlinks/README.md
+dbc91ee09dbf426c3a2b5998f25bed82039df408e639a051d8172e49a2bb06a7  video_symlinks/data.parquet

level3/interleaved_input_list_for_GEM/README.md

@@ -0,0 +1,3 @@
+# interleaved_input_list_for_GEM
+
+See [Level3 release guide](../README.md) for the schema and reconstruction commands. This directory stores portable tables; use the reconstruction script to restore the original file layout.

level3/interleaved_input_list_for_GEM/items.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f4222a67e2f2cdefbdadb3f9ed18dd296835e969e1823a693ed48edea6c2eb9
+size 1610025

level3/release_manifest.json

@@ -0,0 +1,17 @@
+{
+  "schema_version": "1.0",
+  "tasks": 7764,
+  "instructions": 17575,
+  "videos": 8475,
+  "gem_items": 26050,
+  "gem_texts": 17575,
+  "gem_text_mismatches": 0,
+  "mismatch_item_ids": [],
+  "images_included": false,
+  "audio_transcripts_included": false,
+  "video_bytes_included": false,
+  "public_root_layout": "Data_Processed/Level3/Video_Slices/",
+  "video_identity": "source_size + source_sha256",
+  "parquet_roundtrip": "passed",
+  "source_data_modified": false
+}

level3/scripts/materialize_level3.py

@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+"""Restore or verify the three released Level3 assets. Requires pyarrow."""
+import argparse
+import hashlib
+import json
+import os
+import shutil
+from pathlib import Path
+import pyarrow.parquet as pq
+
+def sha(path):
+    h=hashlib.sha256()
+    with path.open('rb') as f:
+        for b in iter(lambda:f.read(4*1024*1024),b''):h.update(b)
+    return h.hexdigest()
+
+def under(root, relative):
+    p=Path(relative)
+    if p.is_absolute() or '..' in p.parts:raise ValueError('Unsafe relative path: '+str(p))
+    return root/p
+
+def main():
+    p=argparse.ArgumentParser(description=__doc__)
+    p.add_argument('--release-root',type=Path,default=Path(__file__).resolve().parents[1])
+    p.add_argument('--public-root',type=Path,required=True,help='Directory containing Data_Processed/Level3/Video_Slices')
+    p.add_argument('--output-root',type=Path,required=True)
+    p.add_argument('--video-mode',choices=['symlink','copy'],default='symlink')
+    p.add_argument('--verify-only',action='store_true')
+    a=p.parse_args();r=a.release_root.resolve();out=a.output_root.resolve();public=a.public_root.resolve()
+    for line in (r/'SHA256SUMS').read_text().splitlines():
+        expected,name=line.split('  ',1)
+        if sha(under(r,name))!=expected:raise SystemExit('Release checksum mismatch: '+name)
+    texts=pq.read_table(r/'text_audio_instructions/data.parquet').to_pylist()
+    videos=pq.read_table(r/'video_symlinks/data.parquet').to_pylist()
+    items=pq.read_table(r/'interleaved_input_list_for_GEM/items.parquet').to_pylist()
+    source_text={x['item_id']:x['text'].encode('utf-8') for x in texts}
+    gemtext=dict(source_text)
+    overrides=r/'interleaved_input_list_for_GEM/text_overrides.parquet'
+    if overrides.exists():gemtext.update({x['item_id']:x['text'].encode('utf-8') for x in pq.read_table(overrides).to_pylist()})
+    video_by_id={x['item_id']:x for x in videos};bad=[]
+    for x in videos:
+        path=under(public,x['source_relpath'])
+        if not path.is_file():bad.append({'item_id':x['item_id'],'reason':'missing','path':x['source_relpath']})
+        elif path.stat().st_size!=x['source_size'] or sha(path)!=x['source_sha256']:
+            bad.append({'item_id':x['item_id'],'reason':'size_or_sha256_mismatch','path':x['source_relpath']})
+    if bad:
+        print(json.dumps({'ok':False,'invalid_source_videos':bad},ensure_ascii=False,indent=2))
+        raise SystemExit(2)
+    # Complete preflight before creating output. Never overwrite different existing content.
+    plans=[]
+    for x in texts:
+        b=source_text[x['item_id']]
+        if hashlib.sha256(b).hexdigest()!=x['sha256']:raise ValueError('Text checksum mismatch')
+        plans.append((under(out/'text_audio_instructions',x['file_name']),b,None))
+    for x in videos:plans.append((under(out/'video_symlinks',x['file_name']),None,under(public,x['source_relpath'])))
+    for x in items:
+        dest=under(out/'interleaved_input_list_for_GEM',x['file_name'])
+        if x['kind']=='text':
+            b=gemtext[x['item_id']]
+            if hashlib.sha256(b).hexdigest()!=x['sha256']:raise ValueError('GEM text checksum mismatch')
+            plans.append((dest,b,None))
+        else:
+            v=video_by_id[x['item_id']]
+            if v['source_relpath']!=x['source_relpath'] or v['source_sha256']!=x['sha256']:raise ValueError('Video reference mismatch')
+            plans.append((dest,None,under(public,v['source_relpath'])))
+    failures=[]
+    for dest,b,source in plans:
+        if dest.exists() or dest.is_symlink():
+            valid=(not dest.is_symlink() and dest.is_file() and dest.read_bytes()==b) if b is not None else ((dest.is_symlink() and dest.resolve()==source.resolve() and dest.is_file()) or (not dest.is_symlink() and dest.is_file() and sha(dest)==sha(source)))
+            if not valid:failures.append(str(dest.relative_to(out)))
+        elif a.verify_only:failures.append(str(dest.relative_to(out)))
+        # Existing symlink parents must not route output outside output-root.
+        if out not in dest.parent.resolve().parents and dest.parent.resolve()!=out:raise ValueError('Output parent escapes output-root')
+    if failures:
+        print(json.dumps({'ok':False,'missing_or_conflicting_outputs':failures},indent=2));raise SystemExit(3)
+    if not a.verify_only:
+        for dest,b,source in plans:
+            if dest.exists() or dest.is_symlink():continue
+            dest.parent.mkdir(parents=True,exist_ok=True)
+            if b is not None:
+                with dest.open('xb') as f:f.write(b)
+            elif a.video_mode=='copy':shutil.copyfile(source,dest)
+            else:dest.symlink_to(source)
+    print(json.dumps({'ok':True,'mode':'verify' if a.verify_only else 'materialize','tasks':len({x['task_id'] for x in items}),'grouped_texts':len(texts),'videos':len(videos),'gem_items':len(items),'expected_output_entries':len(plans),'source_video_sha256':'passed'},indent=2))
+
+if __name__=='__main__':main()

level3/text_audio_instructions/README.md

@@ -0,0 +1,3 @@
+# text_audio_instructions
+
+See [Level3 release guide](../README.md) for the schema and reconstruction commands. This directory stores portable tables; use the reconstruction script to restore the original file layout.

level3/text_audio_instructions/data.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1771747ae54f8c1e19e9973e5971bfe3f89293d63e6983d3a320ccdb562e7605
+size 1840505

level3/validation.json

@@ -0,0 +1,25 @@
+{
+  "checked_date": "2026-09-17",
+  "full_original_layout_verification": {
+    "ok": true,
+    "mode": "verify",
+    "tasks": 7764,
+    "grouped_texts": 17575,
+    "videos": 8475,
+    "gem_items": 26050,
+    "expected_output_entries": 52100,
+    "source_video_sha256": "passed"
+  },
+  "parquet_roundtrip": "passed",
+  "gem_text_vs_grouped_text": "17575 files byte-identical",
+  "reconstruction_functional_tests": [
+    "materialize fixture",
+    "verify fixture",
+    "preserve skipped order indices",
+    "reject changed source bytes",
+    "reject output conflict"
+  ],
+  "private_server_path_audit": "passed",
+  "actual_media_bytes_published": false,
+  "video_decoding_or_motion_quality_checked": false
+}

level3/video_symlinks/README.md

@@ -0,0 +1,3 @@
+# video_symlinks
+
+See [Level3 release guide](../README.md) for the schema and reconstruction commands. This directory stores portable tables; use the reconstruction script to restore the original file layout.

level3/video_symlinks/data.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbc91ee09dbf426c3a2b5998f25bed82039df408e639a051d8172e49a2bb06a7
+size 878507