Add files using upload-large-folder tool

.gitignore

@@ -0,0 +1,3 @@
+__pycache__/
+*.pyc
+.cache/

README.md

@@ -0,0 +1,122 @@
+---
+pretty_name: ESI-Bench Passive Single-View Renders
+language:
+- en
+task_categories:
+- visual-question-answering
+size_categories:
+- 1K<n<10K
+license: other
+tags:
+- embodied-ai
+- spatial-reasoning
+- synthetic-data
+- omnigibson
+- behavior-1k
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train/**
+  - split: validation
+    path: data/validation/**
+---
+
+# ESI-Bench Passive Single-View Renders
+
+This standalone dataset contains one passive RGB observation for successfully rendered ESI-Bench question instances. It is designed for visual question answering and passive-observation baselines for embodied spatial intelligence.
+
+The repository uses the Hugging Face ImageFolder format. Each image is paired with a row in the nearest `metadata.jsonl` through the `file_name` column.
+
+## Dataset Structure
+
+The release contains 3,422 PNG images at 1280 x 720 resolution:
+
+| Split | Images | Meaning |
+| --- | ---: | --- |
+| `train` | 3,189 | Passed the automatic preview-quality checks |
+| `validation` | 233 | Rendered successfully but flagged by the automatic quality checks |
+
+These are quality partitions, not conventional machine-learning train and validation splits. The `validation` split contains 220 images flagged as `very_low_information` and 13 flagged as `constant_image`. Consumers should inspect the `quality_reasons_json` and `usable_preview` fields before deciding whether to include these samples.
+
+Eighty manifest entries without a publishable render result and one orphaned PNG without a successful manifest result are excluded. A sanitized audit is provided under `reports/`.
+
+## Metadata Schema
+
+Each metadata row contains:
+
+| Field | Description |
+| --- | --- |
+| `file_name` | Image path relative to the split directory; used by ImageFolder |
+| `id` | Zero-padded ESI-Bench question ID |
+| `source_index` | Original render-manifest index |
+| `big_task`, `small_task` | ESI-Bench task taxonomy |
+| `runner_task` | Internal benchmark runner family |
+| `scene`, `room` | OmniGibson scene and room identifiers |
+| `question`, `answer` | Question and reference answer associated with the rendered observation |
+| `answer_type` | Answer format such as choice, count, boolean, or freeform |
+| `options_json` | JSON-encoded answer options |
+| `render_status` | `rendered` or `rendered_warning` |
+| `quality_reasons_json` | JSON-encoded automatic quality warnings |
+| `usable_preview` | Automatic preview-usability flag |
+| `width`, `height`, `size_bytes` | Image properties measured during rendering |
+| `camera_pose_json` | JSON-encoded camera pose when available |
+| `camera_info_json` | JSON-encoded task-specific camera selection metadata |
+| `env_postprocess_json` | JSON-encoded environment postprocessing metadata |
+
+Local filesystem paths, temporary paths, process logs, retry traces, and simulator stderr/stdout are intentionally excluded from the public metadata.
+
+## Loading
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("ESI-Bench/esi-bench-passive-single")
+clean = dataset["train"]
+quality_warnings = dataset["validation"]
+```
+
+To keep only records accepted by the automatic quality check:
+
+```python
+clean = dataset["train"].filter(lambda row: row["usable_preview"])
+```
+
+## Task Taxonomy
+
+The dataset covers the ten ESI-Bench categories:
+
+- Action Sequencing
+- Cognitive Mapping
+- Enumerative Perception
+- Metric Comparison
+- Perceptual Grounding
+- Physical Dynamics
+- Physical Structure
+- Spatial Relations
+- Specular Reflection
+- Temporal Understanding
+
+## Intended Use and Limitations
+
+The images are synthetic observations rendered in OmniGibson from ESI-Bench task instances. They are intended for embodied spatial reasoning research, visual question answering evaluation, and passive-observation baselines.
+
+Automatic quality heuristics can produce false positives and false negatives. In particular, visually sparse scenes can be valid for some tasks. The quality-warning split is retained to make this decision transparent rather than silently discarding every flagged render.
+
+The data reflects the environments, assets, rendering pipeline, and task generation procedures used by ESI-Bench and should not be treated as representative of real-world visual distributions.
+
+## Citation
+
+```bibtex
+@inproceedings{hong2026esibench,
+  title     = {{ESI-Bench}: Towards Embodied Spatial Intelligence that Closes the Perception-Action Loop},
+  author    = {Hong, Yining and Liu, Jiageng and Yin, Han and Li, Manling and Guibas, Leonidas and Li, Fei-Fei and Wu, Jiajun and Choi, Yejin},
+  year      = {2026}
+}
+```
+
+This dataset also builds on BEHAVIOR-1K and OmniGibson. Please cite their corresponding publications when using the rendered observations.
+
+## License
+
+The ESI-Bench source code is released under the MIT License. The rendered observations incorporate OmniGibson and BEHAVIOR-1K environments and assets and may be subject to their applicable asset and dataset terms. The repository therefore uses `license: other`; users and maintainers should verify the upstream redistribution terms before public release and use.

reports/excluded_records.jsonl

@@ -0,0 +1,80 @@
+{"source_index":398,"big_task":"Enumerative Perception","small_task":"Counting w Occlusion","scene":"Beechwood_0_garden","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1155,"big_task":"Perceptual Grounding","small_task":"Partial Occlusion","scene":"Benevolence_1_int","room":"dining_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1159,"big_task":"Perceptual Grounding","small_task":"Partial Occlusion","scene":"Ihlen_1_int","room":"bathroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1160,"big_task":"Perceptual Grounding","small_task":"Partial Occlusion","scene":"Ihlen_1_int","room":"kitchen_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1173,"big_task":"Perceptual Grounding","small_task":"Partial Occlusion","scene":"Pomaria_0_int","room":"bedroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1207,"big_task":"Perceptual Grounding","small_task":"Partial Occlusion","scene":"house_single_floor","room":"kitchen_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1218,"big_task":"Perceptual Grounding","small_task":"Partial Occlusion","scene":"office_large","room":"meeting_room_4","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1702,"big_task":"Physical Dynamics","small_task":"Inclined Plane","scene":"Wainscott_0_int","room":"living_room_2","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":1829,"big_task":"Physical Structure","small_task":"Deformable","scene":"Beechwood_0_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2062,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_0_int","room":"living_room_1","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2064,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"playroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2065,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"playroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2066,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"playroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2068,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2069,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2070,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2071,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2072,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2073,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2074,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Beechwood_1_int","room":"television_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2076,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_0_int","room":"bathroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2078,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_0_int","room":"corridor_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2080,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_0_int","room":"empty_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2082,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2083,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2084,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2085,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2086,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2088,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Benevolence_2_int","room":"bedroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2090,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Ihlen_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2091,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Ihlen_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2092,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Ihlen_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2094,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Merom_0_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2096,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Pomaria_1_int","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2098,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Wainscott_0_garden","room":"living_room_1","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2099,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Wainscott_0_garden","room":"living_room_1","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2100,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"Wainscott_0_garden","room":"living_room_1","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2102,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"grocery_store_half_stocked","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2104,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"grocery_store_half_stocked","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2105,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"grocery_store_half_stocked","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2106,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"grocery_store_half_stocked","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2107,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"grocery_store_half_stocked","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2108,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"grocery_store_half_stocked","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2110,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_arch_wood","room":"empty_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2112,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2113,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2114,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2116,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2117,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2118,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2119,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2120,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"hall_conference_large","room":"conference_hall_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2122,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2123,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2124,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2125,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2126,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2127,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2128,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"house_double_floor_lower","room":"living_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2130,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"office_vendor_machine","room":"meeting_room_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2132,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2133,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2134,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2135,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2136,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2137,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2138,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2139,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2140,"big_task":"Physical Structure","small_task":"Rigid Containment","scene":"restaurant_hotel","room":"lobby_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2951,"big_task":"Specular Reflection","small_task":"Correspondence","scene":"Pomaria_0_int","room":"private_office_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2952,"big_task":"Specular Reflection","small_task":"Correspondence","scene":"Pomaria_0_int","room":"private_office_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":2953,"big_task":"Specular Reflection","small_task":"Correspondence","scene":"Pomaria_0_int","room":"private_office_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3055,"big_task":"Specular Reflection","small_task":"Reflection Authoring","scene":"grocery_store_convenience","room":"bathroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3056,"big_task":"Specular Reflection","small_task":"Reflection Authoring","scene":"grocery_store_convenience","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3137,"big_task":"Specular Reflection","small_task":"Spatial Relations","scene":"Pomaria_0_int","room":"private_office_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3138,"big_task":"Specular Reflection","small_task":"Spatial Relations","scene":"Pomaria_0_int","room":"private_office_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3139,"big_task":"Specular Reflection","small_task":"Spatial Relations","scene":"Pomaria_0_int","room":"private_office_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3141,"big_task":"Specular Reflection","small_task":"Spatial Relations","scene":"Pomaria_2_int","room":"bedroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3165,"big_task":"Specular Reflection","small_task":"Spatial Relations","scene":"grocery_store_convenience","room":"bathroom_0","render_status":"failed","reason":"no_publishable_render_result"}
+{"source_index":3166,"big_task":"Specular Reflection","small_task":"Spatial Relations","scene":"grocery_store_convenience","room":"grocery_store_0","render_status":"failed","reason":"no_publishable_render_result"}

reports/export_summary.json

@@ -0,0 +1,24 @@
+{
+  "manifest_rows": 3502,
+  "question_rows": 3502,
+  "published_images": 3422,
+  "splits": {
+    "train": 3189,
+    "validation": 233
+  },
+  "excluded_manifest_rows": 80,
+  "excluded_status_counts": {
+    "failed": 80
+  },
+  "orphan_image_count": 1,
+  "orphan_images": [
+    "Enumerative_Perception/Counting_w_Occlusion/Beechwood_0_garden/living_room_0/0398_q_004.png"
+  ],
+  "warning_reason_counts": {
+    "constant_image": 13,
+    "very_low_information": 220
+  },
+  "image_placement": {
+    "linked": 3422
+  }
+}

scripts/export_passive_single.py

@@ -0,0 +1,256 @@
+#!/usr/bin/env python3
+"""Build the standalone ESI-Bench passive-single Hugging Face dataset repo."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import shutil
+from collections import Counter
+from pathlib import Path
+from typing import Any
+
+
+PROJECT_ROOT = Path(__file__).resolve().parents[2]
+DEFAULT_SOURCE_ROOT = PROJECT_ROOT / "outputs/passive_single"
+DEFAULT_MANIFEST = DEFAULT_SOURCE_ROOT / "all_manifest.jsonl"
+DEFAULT_QUESTIONS = PROJECT_ROOT / "hf_dataset/data/questions.jsonl"
+DEFAULT_OUTPUT_ROOT = PROJECT_ROOT / "hf_passive_single_dataset"
+
+
+def read_jsonl(path: Path) -> list[dict[str, Any]]:
+    rows: list[dict[str, Any]] = []
+    with path.open(encoding="utf-8") as handle:
+        for line_number, line in enumerate(handle, start=1):
+            line = line.strip()
+            if not line:
+                continue
+            try:
+                value = json.loads(line)
+            except json.JSONDecodeError as exc:
+                raise SystemExit(f"Invalid JSON in {path}:{line_number}: {exc}") from exc
+            if not isinstance(value, dict):
+                raise SystemExit(f"Expected an object in {path}:{line_number}")
+            rows.append(value)
+    return rows
+
+
+def json_text(value: Any) -> str:
+    return json.dumps(value, ensure_ascii=False, separators=(",", ":"))
+
+
+def relative_image_path(raw_path: str) -> Path:
+    prefix = "outputs/passive_single/"
+    normalized = raw_path.replace("\\", "/")
+    if normalized.startswith(prefix):
+        normalized = normalized[len(prefix) :]
+    relative = Path(normalized)
+    if relative.is_absolute() or ".." in relative.parts:
+        raise SystemExit(f"Unsafe output_image path: {raw_path}")
+    return relative
+
+
+def action_qa(row: dict[str, Any]) -> tuple[str | None, str | None, str | None, str | None]:
+    result = row.get("result")
+    if not isinstance(result, dict):
+        return None, None, None, None
+    camera_info = result.get("camera_info")
+    if not isinstance(camera_info, dict):
+        return None, None, None, None
+    qa = camera_info.get("qa")
+    if not isinstance(qa, dict) or not qa.get("question"):
+        return None, None, None, None
+
+    answers = {
+        key: qa[key]
+        for key in ("answer_A", "answer_C")
+        if key in qa
+    }
+    options: dict[str, Any] = {}
+    if "choices_A" in qa:
+        options["A"] = qa["choices_A"]
+    if "choices_C" in qa:
+        options["C"] = qa["choices_C"]
+    return (
+        str(qa["question"]),
+        json_text(answers) if answers else None,
+        "choice" if answers else None,
+        json_text(options) if options else None,
+    )
+
+
+def public_metadata(
+    manifest_row: dict[str, Any],
+    question_row: dict[str, Any],
+    file_name: str,
+) -> dict[str, Any]:
+    result = manifest_row["result"]
+    quality = result.get("quality") if isinstance(result.get("quality"), dict) else {}
+
+    rendered_question, rendered_answer, rendered_answer_type, rendered_options = action_qa(manifest_row)
+    camera_pose = result.get("camera_pose")
+    camera_info = result.get("camera_info")
+    env_postprocess = result.get("env_postprocess")
+
+    return {
+        "file_name": file_name,
+        "id": str(result["id"]),
+        "source_index": manifest_row.get("index"),
+        "big_task": manifest_row.get("big_task") or result.get("big_task"),
+        "small_task": manifest_row.get("small_task") or result.get("small_task"),
+        "runner_task": question_row.get("runner_task") or result.get("runner_task"),
+        "scene": manifest_row.get("scene") or result.get("scene"),
+        "room": manifest_row.get("room") or result.get("room"),
+        "question": rendered_question or question_row.get("question"),
+        "answer": rendered_answer or question_row.get("answer"),
+        "answer_type": rendered_answer_type or question_row.get("answer_type"),
+        "options_json": rendered_options or question_row.get("options_json") or "{}",
+        "render_status": manifest_row.get("status"),
+        "quality_reasons_json": json_text(quality.get("quality_reasons") or []),
+        "usable_preview": bool(quality.get("usable_preview")),
+        "width": quality.get("width"),
+        "height": quality.get("height"),
+        "size_bytes": quality.get("size_bytes"),
+        "camera_pose_json": json_text(camera_pose or {}),
+        "camera_info_json": json_text(camera_info or {}),
+        "env_postprocess_json": json_text(env_postprocess or {}),
+    }
+
+
+def place_image(source: Path, destination: Path, mode: str) -> str:
+    destination.parent.mkdir(parents=True, exist_ok=True)
+    if mode == "copy":
+        shutil.copy2(source, destination)
+        return "copied"
+    try:
+        os.link(source, destination)
+        return "linked"
+    except OSError:
+        shutil.copy2(source, destination)
+        return "copied"
+
+
+def write_jsonl(path: Path, rows: list[dict[str, Any]]) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    with path.open("w", encoding="utf-8") as handle:
+        for row in rows:
+            handle.write(json_text(row))
+            handle.write("\n")
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--source-root", type=Path, default=DEFAULT_SOURCE_ROOT)
+    parser.add_argument("--manifest", type=Path, default=DEFAULT_MANIFEST)
+    parser.add_argument("--questions", type=Path, default=DEFAULT_QUESTIONS)
+    parser.add_argument("--output-root", type=Path, default=DEFAULT_OUTPUT_ROOT)
+    parser.add_argument("--copy", action="store_true", help="Copy images instead of using hard links when possible.")
+    parser.add_argument("--overwrite", action="store_true", help="Replace an existing generated data directory.")
+    args = parser.parse_args()
+
+    source_root = args.source_root.resolve()
+    manifest_path = args.manifest.resolve()
+    questions_path = args.questions.resolve()
+    output_root = args.output_root.resolve()
+    data_root = output_root / "data"
+    reports_root = output_root / "reports"
+
+    if output_root == PROJECT_ROOT or output_root == source_root:
+        raise SystemExit(f"Refusing unsafe output root: {output_root}")
+    if data_root.exists() or reports_root.exists():
+        if not args.overwrite:
+            raise SystemExit("Generated data/reports already exist. Pass --overwrite to rebuild them.")
+        if data_root.exists():
+            shutil.rmtree(data_root)
+        if reports_root.exists():
+            shutil.rmtree(reports_root)
+
+    manifest_rows = read_jsonl(manifest_path)
+    question_rows = read_jsonl(questions_path)
+    questions_by_id = {str(row["id"]): row for row in question_rows}
+    if len(questions_by_id) != len(question_rows):
+        raise SystemExit("Question IDs are not unique")
+
+    metadata_by_split: dict[str, list[dict[str, Any]]] = {"train": [], "validation": []}
+    excluded: list[dict[str, Any]] = []
+    referenced_images: set[Path] = set()
+    placement_counts: Counter[str] = Counter()
+
+    for row in manifest_rows:
+        status = row.get("status")
+        result = row.get("result")
+        if status not in {"rendered", "rendered_warning"} or not isinstance(result, dict):
+            excluded.append(
+                {
+                    "source_index": row.get("index"),
+                    "big_task": row.get("big_task"),
+                    "small_task": row.get("small_task"),
+                    "scene": row.get("scene"),
+                    "room": row.get("room"),
+                    "render_status": status,
+                    "reason": "no_publishable_render_result",
+                }
+            )
+            continue
+
+        item_id = str(result.get("id") or "")
+        if not item_id or item_id not in questions_by_id:
+            raise SystemExit(f"Missing standardized question metadata for rendered ID {item_id!r}")
+        raw_image_path = result.get("output_image")
+        if not isinstance(raw_image_path, str):
+            raise SystemExit(f"Rendered ID {item_id} has no output_image")
+
+        relative = relative_image_path(raw_image_path)
+        source = source_root / relative
+        if not source.is_file():
+            raise SystemExit(f"Missing rendered image for ID {item_id}: {source}")
+        referenced_images.add(relative)
+
+        split = "train" if status == "rendered" else "validation"
+        destination = data_root / split / relative
+        placement_counts[place_image(source, destination, "copy" if args.copy else "link")] += 1
+        metadata_by_split[split].append(
+            public_metadata(row, questions_by_id[item_id], relative.as_posix())
+        )
+
+    for split, rows in metadata_by_split.items():
+        rows.sort(key=lambda row: row["id"])
+        write_jsonl(data_root / split / "metadata.jsonl", rows)
+
+    all_pngs = {
+        path.relative_to(source_root)
+        for path in source_root.rglob("*.png")
+        if path.is_file()
+    }
+    orphan_images = sorted(path.as_posix() for path in all_pngs - referenced_images)
+    failed_counts = Counter(row.get("render_status") for row in excluded)
+    warning_reasons = Counter(
+        reason
+        for row in metadata_by_split["validation"]
+        for reason in json.loads(row["quality_reasons_json"])
+    )
+
+    summary = {
+        "manifest_rows": len(manifest_rows),
+        "question_rows": len(question_rows),
+        "published_images": sum(len(rows) for rows in metadata_by_split.values()),
+        "splits": {split: len(rows) for split, rows in metadata_by_split.items()},
+        "excluded_manifest_rows": len(excluded),
+        "excluded_status_counts": dict(sorted(failed_counts.items(), key=lambda item: str(item[0]))),
+        "orphan_image_count": len(orphan_images),
+        "orphan_images": orphan_images,
+        "warning_reason_counts": dict(sorted(warning_reasons.items())),
+        "image_placement": dict(sorted(placement_counts.items())),
+    }
+    write_jsonl(reports_root / "excluded_records.jsonl", excluded)
+    (reports_root / "export_summary.json").write_text(
+        json.dumps(summary, ensure_ascii=False, indent=2) + "\n",
+        encoding="utf-8",
+    )
+    print(json.dumps(summary, ensure_ascii=False, indent=2))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

scripts/upload_hf_dataset.py

@@ -0,0 +1,68 @@
+#!/usr/bin/env python3
+"""Create and upload the standalone passive-single dataset repository."""
+
+from __future__ import annotations
+
+import argparse
+import json
+from pathlib import Path
+
+from huggingface_hub import HfApi
+
+
+DEFAULT_REPO_ROOT = Path(__file__).resolve().parents[1]
+
+
+def validate(repo_root: Path) -> None:
+    required = [
+        repo_root / "README.md",
+        repo_root / "data/train/metadata.jsonl",
+        repo_root / "data/validation/metadata.jsonl",
+        repo_root / "reports/export_summary.json",
+    ]
+    missing = [str(path) for path in required if not path.is_file()]
+    if missing:
+        raise SystemExit("Missing required files:\n" + "\n".join(missing))
+
+    summary = json.loads((repo_root / "reports/export_summary.json").read_text(encoding="utf-8"))
+    if summary.get("published_images") != 3422:
+        raise SystemExit(f"Unexpected published image count: {summary.get('published_images')}")
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--repo-id", default="ESI-Bench/esi-bench-passive-single")
+    parser.add_argument("--repo-root", type=Path, default=DEFAULT_REPO_ROOT)
+    parser.add_argument("--private", action="store_true")
+    parser.add_argument("--revision", default=None)
+    parser.add_argument("--num-workers", type=int, default=None)
+    args = parser.parse_args()
+
+    repo_root = args.repo_root.resolve()
+    validate(repo_root)
+
+    api = HfApi()
+    api.create_repo(
+        repo_id=args.repo_id,
+        repo_type="dataset",
+        private=args.private,
+        exist_ok=True,
+    )
+    api.upload_large_folder(
+        repo_id=args.repo_id,
+        repo_type="dataset",
+        folder_path=repo_root,
+        revision=args.revision,
+        num_workers=args.num_workers,
+        ignore_patterns=[
+            "**/__pycache__/**",
+            "**/*.pyc",
+            ".git/**",
+        ],
+    )
+    print(f"Uploaded to https://huggingface.co/datasets/{args.repo_id}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())