bugsinpy_v2: refreshed release with current v1 tasks, uniform card and trace statistics

bugsinpy_v2/README.md

@@ -2,39 +2,20 @@
 
 ## Dataset stats
 
-- **489 v1 tasks → 65 kept; 424 newly archived.** All 65 retain their v1 repairs; v2 changes no task instructions, environments, tests or reference solutions.
-- **501 native bugs: 65 kept, 435 archived and 1 not converted.** Archives include the 11 earlier v1 exclusions.
-- **Images:** the original benchmark supplied none; v1 introduced 11 shared images. The 65 kept tasks reuse **4** of those [cached images](https://huggingface.co/datasets/FWeindel/validated-tasks/tree/main/bugsinpy_v1/cached_images). The Parquet pins their immutable revisions and checksums; no image copies are needed in v2.
-
-## Dataset characteristics
-
-Real Python projects at buggy commits, with upstream regression tests and reference fixes.
-
-<details><summary>Domains</summary>
-
-Software Engineering; Debugging; Python repositories.
-
-</details>
-
-<details><summary>Capabilities</summary>
-
-Repository exploration, error diagnosis, code repair and regression testing.
-
-</details>
-
-**Likely transfer targets:** repository debugging and SWE-bench-style tasks.
+- **v1 tasks evaluated:** 489
+- **Kept:** 65 (solved at least once by either model)
+- **Excluded:** 424 (no recorded success)
+- **Attempts:** pass@16 with Qwen3-30B Instruct and Qwen3-30B Thinking (15,648 in total)
 
 ## Pass@16 filtering
 
-Keep a task if either model solved it at least once. Both models completed 16 attempts on every v1 task. Terminal errors count as unsuccessful attempts; unsolved tasks are not necessarily defective.
-
 | Model | Tasks evaluated | Attempts | Tasks solved at least once |
 | --- | ---: | ---: | ---: |
 | Qwen3-30B Instruct | 489 | 7,824 | 56 |
 | Qwen3-30B Thinking | 489 | 7,824 | 47 |
-| Either model | 489 | 15,648 | **65** |
+| Either model | 489 | 15,648 | 65 |
 
-Successful-attempt distribution **among the 65 kept tasks**. Each model column totals 65; rows with zero tasks for both models are omitted. A task with 0/16 for one model was solved by the other.
+### Successful attempts among kept tasks
 
 | Successful attempts | Instruct tasks | Thinking tasks |
 | --- | ---: | ---: |
@@ -54,50 +35,41 @@ Successful-attempt distribution **among the 65 kept tasks**. Each model column t
 | 14/16 | 0 | 1 |
 | 15/16 | 1 | 0 |
 
-The kept tasks inherit v1's rule-based repairs, build checks, reference reward 1 and no-op reward 0. [V1 repair details](https://huggingface.co/datasets/FWeindel/validated-tasks/blob/main/bugsinpy_v1/README.md) · [Teacher results](https://huggingface.co/datasets/FWeindel/validated-tasks/discussions/7).
-
 ## Select tasks by successful-attempt count
 
-`tasks.parquet` includes `instruct_successes`, `thinking_successes` and `attempts_per_model` (16). Counts are also available in [selection.json](metadata/selection.json). Excluded payloads remain in the pinned v1 source recorded in `selection.json`; no duplicate archive Parquet is uploaded.
+You can select tasks by how often they were solved, for example all tasks Instruct solved exactly once:
 
 ```python
 import pyarrow.parquet as pq
 
-# Exactly 1 successful attempt out of 16 for Instruct:
-filters = [("instruct_successes", "=", 1)]
+subset = pq.read_table("tasks.parquet", filters=[("instruct_successes", "=", 1)])
+```
+
+## Traces
+
+`traces/` holds all 15,648 recorded attempts, successful and failed, from every evaluated task, one row per attempt (Zstandard-compressed Parquet). Columns include `task_id`, `model`, `attempt_id`, `reward`, `verifier_reward`, `retained_v2` and the full `trajectory` and `result` JSON.
 
-# Alternatively, at least 3 successes for either model:
-# filters = [[("instruct_successes", ">=", 3)],
-#            [("thinking_successes", ">=", 3)]]
+```python
+import json
+import pyarrow.dataset as ds
 
-subset = pq.read_table("tasks.parquet", filters=filters)
-pq.write_table(subset, "selected.parquet")
+traces = ds.dataset("traces", format="parquet")
+failed = traces.to_table(filter=ds.field("reward") == 0)
+conversation = json.loads(failed["trajectory"][0].as_py())
 ```
 
-`patch.py` reproduces the release from the pinned v1 Parquets and recorded counts.
+Trajectory statistics (plots, command distribution) are in [trace_stats](trace_stats/README.md).
 
-## Runtime and validation time
+Not yet checked with the repository-leak checks added on 2026-10-17 (leftover git history and file timestamps that could reveal the solution).
+
+## Runtime
 
 - **Versions:** Apptainer **1.5.4**, Python **3.12.14**, Harbor **0.8.1**.
 - **Implementation:** [Marin Harbor `7faf878c14b6`](https://github.com/marin-community/harbor/tree/7faf878c14b6d72737579ec936ebf5f74ba3194d) + [our runtime patches](https://github.com/MLI-lab/OT-next-data/tree/main/harbor_patches).
 
-| Teacher | Mean time per attempt | Median time per attempt | Attempts measured |
+| Model | Mean time per attempt | Median time per attempt | Attempts measured |
 | --- | ---: | ---: | ---: |
 | Instruct | 819.81 s | 591.60 s | 7,824 |
-| Thinking | 1098.50 s | 750.80 s | 7,824 |
-
-Timings cover recorded trials across all 489 evaluated tasks, including unsuccessful attempts.
-
-## Traces
+| Thinking | 1,098.50 s | 750.80 s | 7,824 |
 
-[Trace Parquets](traces) contain **15,648 attempts**, including successful and failed attempts from every evaluated task, kept or excluded. Zstandard-compressed, one row per attempt; **0** rows have no saved conversation and are marked `trajectory_found=false`.
-
-Columns include `task_id`, `model`, `attempt_id`, `reward`, `verifier_reward`, `retained_v2`, `trajectory_found`, and the full `trajectory` and `result` JSON. `reward` follows the published evaluation policy; `verifier_reward` preserves the raw verifier score. The manifest pins the original trace source.
-
-```python
-import json
-import pyarrow.dataset as ds
-traces = ds.dataset("traces", format="parquet", exclude_invalid_files=True)
-failed = traces.to_table(filter=(ds.field("reward") == 0) & ds.field("trajectory_found"))
-conversation = json.loads(failed["trajectory"][0].as_py())
-```
+`patch.py` rebuilds `tasks.parquet` from the current v1 tasks and the recorded results, both pinned to revision `a43f5635`.

bugsinpy_v2/patch.py

@@ -1,125 +1,86 @@
-"""Reproduce BugsInPy v2 from pinned v1 Parquets and the recorded pass@16 results.
+"""Reproduce BugsInPy v2 tasks.parquet: the current v1 payload of every kept task plus the recorded results.
 
 pip install pyarrow huggingface_hub
 python patch.py --out ./rebuilt
-Use --source-dir PATH to reuse the two original v1 Parquets locally.
+
+Both inputs are pinned to repository revision a43f5635 and checked by SHA-256:
+- bugsinpy_v1/tasks.parquet: the current v1 payloads, content hashes and image manifest.
+- bugsinpy_v2/tasks.parquet as published before this refresh: which tasks are kept, the
+  recorded success counts (`instruct_successes`, `thinking_successes`, `attempts_per_model`) and the release metadata.
+Use --source-dir PATH (holding v1.parquet and v2.parquet) to reuse local copies.
+Trajectories and success counts stay as recorded; each trace row keeps the content hash it
+was evaluated on.
 """
 import argparse
 import copy
 import hashlib
 import json
 from pathlib import Path
+
 import pyarrow as pa
 import pyarrow.parquet as pq
 
-MODELS = {'instruct_successes': 'qwen3-30b-instruct-2507',
-          'thinking_successes': 'qwen3-30b-thinking'}
+REPO = 'FWeindel/validated-tasks'
+REVISION = 'a43f56353751d6d73a93ff448de6981040c4d0c5'
+SOURCES = {'v1': ('bugsinpy_v1/tasks.parquet', 'ca7d89da898adc14f5ce547b49552197d726010e89282ffe30024bda949a636c'),
+           'v2': ('bugsinpy_v2/tasks.parquet', '4b066529e9e31959ee70af980471c3e3fd4268a32b58aac98f4eda2cef56638b')}
+REFRESHED = '2026-10-17'
+
+
+def source(name, source_dir, cache):
+    path_in_repo, digest = SOURCES[name]
+    if source_dir is None:
+        from huggingface_hub import hf_hub_download
+        path = Path(hf_hub_download(REPO, path_in_repo, repo_type='dataset', revision=REVISION, cache_dir=cache))
+    else:
+        path = source_dir / f'{name}.parquet'
+    if hashlib.sha256(path.read_bytes()).hexdigest() != digest:
+        raise ValueError(f'pinned source hash mismatch: {path_in_repo}')
+    return path
 
-def rows(path):
-    table = pq.ParquetFile(path)
-    for i in range(table.num_row_groups):
-        yield from table.read_row_group(i).to_pylist()
 
-def restrict_images(manifest, names):
+def restrict_images(manifest, kept):
     result = copy.deepcopy(manifest)
-    result['tasks'] = {k: v for k, v in manifest['tasks'].items() if k in names}
-    bundles = {e['bundle'] for t in result['tasks'].values() for e in t['environments']}
-    result['bundles'] = {k: v for k, v in manifest['bundles'].items() if k in bundles}
-    result['missing'] = []
+    result['tasks'] = {path: entry for path, entry in manifest['tasks'].items() if path in kept}
+    used = {env['bundle'] for entry in result['tasks'].values() for env in entry['environments']}
+    result['bundles'] = {bid: bundle for bid, bundle in manifest['bundles'].items() if bid in used}
+    result['missing'] = [m for m in manifest.get('missing', []) if m.get('task') in kept]
     return result
 
-def build(source_dir, out, selection_path):
-    selection = json.loads(selection_path.read_text())
+
+def build(source_dir, out):
     out.mkdir(parents=True, exist_ok=True)
-    if any((out / n).exists() for n in ('tasks.parquet', 'archive.parquet')):
-        raise ValueError('Output Parquets already exist; choose an empty output directory')
-    sources = {}
-    for name in ('tasks.parquet', 'archive.parquet'):
-        if source_dir is None:
-            from huggingface_hub import hf_hub_download
-            path = Path(hf_hub_download(selection['repo_id'], selection['source_folder'] + '/' + name,
-                        repo_type='dataset', revision=selection['source_revision'], cache_dir=out / '.source-cache'))
-        else:
-            path = source_dir / name
-        with path.open('rb') as handle:
-            digest = hashlib.file_digest(handle, 'sha256').hexdigest()
-        if digest != selection['source_sha256'][name]:
-            raise ValueError('Pinned source hash mismatch: ' + name)
-        sources[name] = path
-    tasks = selection['tasks']
-    kept = {name for name, task in tasks.items() if task['kept']}
-    assert len(tasks) == 489 and len(kept) == 65
-    original_schema = pq.read_schema(sources['tasks.parquet'])
-    image_manifest = json.loads(original_schema.metadata[b'ot.images.v1'])
-    images = restrict_images(image_manifest, kept)
-    assert set(images['tasks']) == kept
-    release = {k: v for k, v in selection.items() if k != 'tasks'}
-    release['selection_file'] = 'metadata/selection.json'
-    metadata = {b'ot.release.v2': json.dumps(release).encode(), b'ot.images.v1': json.dumps(images).encode()}
-    schema = original_schema.remove_metadata().append(pa.field('instruct_successes', pa.int64())).append(
-        pa.field('thinking_successes', pa.int64())).append(pa.field('attempts_per_model', pa.int64()))
-    seen, audit = set(), {}
-    with pq.ParquetWriter(out / 'tasks.parquet', schema.with_metadata(metadata), compression='zstd') as retain, \
-         pq.ParquetWriter(out / 'archive.parquet', schema.with_metadata({b'ot.release.v2': json.dumps(release).encode()}), compression='zstd') as archive:
-        counts = {'retained': 0, 'newly_archived': 0, 'prior_archived': 0}
-        for row in rows(sources['tasks.parquet']):
-            name = row['path']
-            if name in seen or name not in tasks:
-                raise ValueError('Duplicate or unexpected task: ' + name)
-            seen.add(name)
-            result = tasks[name]
-            assert row['content_sha256'] == result['content_sha256']
-            for column, model in MODELS.items():
-                rewards = result['models'][model]['rewards']
-                assert len(rewards) == 16 and len(set(result['models'][model]['attempt_ids'])) == 16
-                row[column] = sum(reward >= 1 for reward in rewards)
-                assert row[column] == result['models'][model]['solved_attempts']
-            row['attempts_per_model'] = 16
-            selected = row['instruct_successes'] > 0 or row['thinking_successes'] > 0
-            assert selected == result['kept']
-            audit[name] = {'binary_sha256': hashlib.sha256(row['task_binary']).hexdigest(),
-                           'content_sha256': row['content_sha256'], 'kept': selected}
-            if selected:
-                row['stages_passed'] = ','.join(dict.fromkeys(row['stages_passed'].split(',') + ['6']))
-                retain.write_table(pa.Table.from_pylist([row], schema=schema.with_metadata(metadata)))
-                counts['retained'] += 1
-            else:
-                row['archive_stage'] = 6
-                row['archive_reason'] = 'pass16-unsolved: no successful attempt in 16 Instruct and 16 Thinking trials; this is not a confirmed task defect.'
-                row['archive_labels'] = list(row.get('archive_labels') or []) + ['pass16-unsolved']
-                archive.write_table(pa.Table.from_pylist([row], schema=schema))
-                counts['newly_archived'] += 1
-        assert seen == set(tasks)
-        for row in rows(sources['archive.parquet']):
-            assert row['path'] not in seen
-            seen.add(row['path'])
-            for column in MODELS:
-                row[column] = None
-            row['attempts_per_model'] = None
-            audit[row['path']] = {'binary_sha256': hashlib.sha256(row['task_binary']).hexdigest(),
-                                  'content_sha256': row['content_sha256'], 'kept': False}
-            archive.write_table(pa.Table.from_pylist([row], schema=schema))
-            counts['prior_archived'] += 1
-    assert counts == {'retained': 65, 'newly_archived': 424, 'prior_archived': 11}
-    checked = set()
-    for filename, expected_kept in [('tasks.parquet', True), ('archive.parquet', False)]:
-        for row in rows(out / filename):
-            prior = audit[row['path']]
-            assert row['path'] not in checked and prior['kept'] == expected_kept
-            assert hashlib.sha256(row['task_binary']).hexdigest() == prior['binary_sha256']
-            assert row['content_sha256'] == prior['content_sha256']
-            checked.add(row['path'])
-    assert checked == seen
-    print(json.dumps(dict(counts, preserved_task_payloads=len(checked), cached_bundles_reused=len(images['bundles']))))
+    if (out / 'tasks.parquet').exists():
+        raise ValueError('tasks.parquet already exists; choose an empty output directory')
+    v1 = pq.read_table(source('v1', source_dir, out / '.source-cache'))
+    old = pq.read_table(source('v2', source_dir, out / '.source-cache'))
+    current = {row['path']: row for row in v1.to_pylist()}
+    extra = [name for name in old.schema.names if name not in v1.schema.names]
+    rows, earlier = [], 0
+    for row in old.to_pylist():
+        fresh = dict(current[row['path']])
+        fresh.update({name: row[name] for name in extra})
+        # The published v2 hash is the one the teacher runs were scored on.
+        earlier += fresh['content_sha256'] != row['content_sha256']
+        rows.append(fresh)
+    schema = pa.schema([*v1.schema, *(old.schema.field(name) for name in extra)])
+    metadata = dict(old.schema.metadata or {})
+    if b'ot.images.v1' in (v1.schema.metadata or {}):
+        images = restrict_images(json.loads(v1.schema.metadata[b'ot.images.v1']), {r['path'] for r in rows})
+        metadata[b'ot.images.v1'] = json.dumps(images).encode()
+    release = json.loads(metadata.get(b'ot.release.v2', b'{}'))
+    release['current_v1'] = {'folder': 'bugsinpy_v1', 'revision': REVISION, 'sha256': SOURCES['v1'][1],
+                             'refreshed': REFRESHED, 'kept_tasks_evaluated_on_earlier_version': earlier}
+    metadata[b'ot.release.v2'] = json.dumps(release).encode()
+    pq.write_table(pa.Table.from_pylist(rows, schema=schema).replace_schema_metadata(metadata),
+                   out / 'tasks.parquet', compression='zstd')
+    print(json.dumps({'kept': len(rows), 'evaluated_on_earlier_version': earlier,
+                      'images': b'ot.images.v1' in metadata}))
+
 
 if __name__ == '__main__':
-    parser = argparse.ArgumentParser(description=__doc__)
+    parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
     parser.add_argument('--source-dir', type=Path)
     parser.add_argument('--out', type=Path, required=True)
-    parser.add_argument('--selection', type=Path, default=Path(__file__).parent / 'metadata' / 'selection.json')
-    parser.add_argument('--include-archive', action='store_true', help='Also reconstruct excluded payloads locally')
     args = parser.parse_args()
-    build(args.source_dir, args.out, args.selection)
-
-    if not args.include_archive:
-        (args.out / 'archive.parquet').unlink()
+    build(args.source_dir, args.out)

bugsinpy_v2/tasks.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4b066529e9e31959ee70af980471c3e3fd4268a32b58aac98f4eda2cef56638b
-size 81801450
+oid sha256:36a62f7da90d7e0b2d86e737acf1975891ca624b59bcc82ee21f97b6d8f799f4
+size 81550488

bugsinpy_v2/trace_stats/README.md

@@ -0,0 +1,34 @@
+# Trajectory statistics
+
+15,648 attempts across 489 tasks of bugsinpy_v2, kept and excluded, each counted once. Task success counts are in the [dataset overview](../README.md).
+
+| Metric | qwen3-30b-instruct-2507 | qwen3-30b-thinking |
+| --- | ---: | ---: |
+| Recorded attempts | 7,824 | 7,824 |
+| Median agent turns | 35 | 32 |
+| Median input tokens per attempt | 462,749 | 383,801 |
+| Median output tokens per attempt | 21,119 | 35,764 |
+| Median agent runtime (seconds) | 346 | 500 |
+| Median verifier runtime (seconds) | 18 | 17 |
+
+Medians use recorded measurements only.
+
+## Plots
+
+![Trace measurements](plots/all-models-trace-distributions.png)
+
+![Trace outcomes](plots/all-models-trace-outcomes.png)
+
+![Most frequent commands](plots/all-models-trace-commands.png)
+
+Commands are counted as submitted, not as successful executions.
+
+Per-model plots: [qwen3-30b-instruct-2507](plots/qwen3-30b-instruct-2507-trace-commands.png), [qwen3-30b-instruct-2507](plots/qwen3-30b-instruct-2507-trace-distributions.png), [qwen3-30b-instruct-2507](plots/qwen3-30b-instruct-2507-trace-outcomes.png), [qwen3-30b-thinking](plots/qwen3-30b-thinking-trace-commands.png), [qwen3-30b-thinking](plots/qwen3-30b-thinking-trace-distributions.png), [qwen3-30b-thinking](plots/qwen3-30b-thinking-trace-outcomes.png).
+
+## Data
+
+- [Summary](data/summary.json): overall and per-model statistics.
+- [Per-task metrics](data/per_task.parquet) and [per-attempt metrics](data/per_attempt.parquet), with nested metrics stored as JSON.
+- [Provenance](data/provenance.json): source files and deduplication.
+
+Attempts without a saved conversation: 0; without a verifier reward: 2,708. Their missing measurements stay empty.

bugsinpy_v2/trace_stats/data/per_attempt.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b22a9ed9034f811021050b15475da3be88707569e73b950cfee9379b817cf418
+size 2503848

bugsinpy_v2/trace_stats/data/per_task.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d2e23800ce2bfba7abbd5b18ca3d4db6c38eaa955893b6fbd51ee65ab836cec
+size 685801

bugsinpy_v2/trace_stats/data/provenance.json

@@ -0,0 +1,548 @@
+{
+ "deduplication_key": [
+  "model",
+  "task_id",
+  "attempt_id"
+ ],
+ "sources": [
+  {
+   "file": "traces/000.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "ac5ab9fec8a71d23387a9ae95c67f57700deaa3de4710658b027705a4ce72f9d"
+  },
+  {
+   "file": "traces/001.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "3f86bff959446f6e491d71f5b0b3a41e9d8d02613e01383c738c3d042501ef54"
+  },
+  {
+   "file": "traces/002.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "343f322fe22fee0f9cc253f99d4b27d1415f278f226a36d6f7a4d51a255b82dd"
+  },
+  {
+   "file": "traces/003.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "d9cf5123d76570b500d48b920c4123eaf345fee027b7aa3679e11931a5aa482b"
+  },
+  {
+   "file": "traces/004.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "0eef881db55d4b21a6d68da7f57295e215c4e8e93784e22e273d5c7f767a5edd"
+  },
+  {
+   "file": "traces/005.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "0087db61ed674d22bad82d5cee240d5e0096fa00b20ec4a2b09d93a5f41d0a81"
+  },
+  {
+   "file": "traces/006.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "f0ad5edab970ef960eeafa17aef19199045d6487d8c5ae1e929db73f550bf763"
+  },
+  {
+   "file": "traces/007.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "b74e3b935403ff4068b7ab1bbfb155cdfb4415fce97a7a8bfacb032e22c82fd9"
+  },
+  {
+   "file": "traces/008.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "c2139d92f0bc2831bdb94b8fcd28975a9d4d4889f87fa14377a8fc28fc630b30"
+  },
+  {
+   "file": "traces/009.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "b733f735937356361cd06b52abd14c89aeb35c5516c843e85c1722f8d84f3556"
+  },
+  {
+   "file": "traces/010.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "eb82dff3c50854a75f0542f3e43c2651863f64cc7b89094a722ee67e7b3320c4"
+  },
+  {
+   "file": "traces/011.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "0cdd1136ded946705ee4e7476a20f3e57d97994c7f23de90c8bd0b7c9ab553cb"
+  },
+  {
+   "file": "traces/012.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "f30c6570f38d0c5d30662f017c97cabecbfb9937ca9b4b1dcc8addf2873b87e6"
+  },
+  {
+   "file": "traces/013.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "9b6d2e69e19ff098e246fcfd4f27316085fd37dff12d1b048ec33952294c1b79"
+  },
+  {
+   "file": "traces/014.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "12f9d28ee9b4e6bcd245b1c13c971b3bd31a0d8690ab6ba09b5abac335e0785a"
+  },
+  {
+   "file": "traces/015.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "97f60576657243dbdad7184160cf2486dbce3ee2a85bd4a5c258643563e2ddb8"
+  },
+  {
+   "file": "traces/016.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "12344193fce39c0512787ceedbe524ff2898f872191df4720d898ba698da2bb1"
+  },
+  {
+   "file": "traces/017.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "72b1a341628c81748a5f21164a8cf686ac10751a3ebdcab1182d6a521a72410f"
+  },
+  {
+   "file": "traces/018.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "18809521ecf57011c007dad65ded6a89c6685fd2128746bc5914f07d33cc118b"
+  },
+  {
+   "file": "traces/019.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "4b0f8b8341925d8d42d26e1b7bf5840370ddfcb01dc89476497a2f42b90343f8"
+  },
+  {
+   "file": "traces/020.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "cafc5af8adccfe44e144233f17d1a01f5ab1e4b08b29b59e2205c177a8a08ff7"
+  },
+  {
+   "file": "traces/021.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "1bf08da225488908b09e5a8e0b77aa1c5717c6108a1d29f4c1bb223d31e5a669"
+  },
+  {
+   "file": "traces/022.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "0f491a103d1b6277fd4f3d0891aa10904bae482e94bd592a1ff0985b97842d3f"
+  },
+  {
+   "file": "traces/023.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "8bd43dcd61389fa5fe8c8614cdc137ec44916ce8dc95012f86acb00364c9b3ca"
+  },
+  {
+   "file": "traces/024.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "24f5195fb30c3b755e50fadc96d1a242b53469c2293117f37dfa445b50e5e965"
+  },
+  {
+   "file": "traces/025.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "745d01d4b228d2de932ab7ca165ddd7decb206f7a2d5381f778bebd274bde37d"
+  },
+  {
+   "file": "traces/026.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "be468d5c1fcfee80b8cca43d2ea004d4836038ef1c90999be7236e72d247e169"
+  },
+  {
+   "file": "traces/027.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "b01973aa694e49d49353256acd0e708a3eb68a2d9e1b3cfc45559d12839e0d72"
+  },
+  {
+   "file": "traces/028.parquet",
+   "rows": 16,
+   "unique_rows_added": 16,
+   "sha256": "1955ee49f1fa052cf2590dea187ef7f2dfe773686c759659375d08178b54049f"
+  },
+  {
+   "file": "traces/029.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "4a82ada9bbbe390eb6ca3fb094c30c7a66e08b1f0ac52b5505ee687397d5fe27"
+  },
+  {
+   "file": "traces/030.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "a6ead8a13a7c4411d75e5176aca5b56ff8fa9373770584e88d78933890d875d2"
+  },
+  {
+   "file": "traces/031.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "6a7f6d88d1f946de71f291df729c8495e7b833789e8bc6cfe18fa24a61f48041"
+  },
+  {
+   "file": "traces/032.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "6bf06454949b3f766f7dbf43fee90606bccfd9f55a41db44b50909293a5154d8"
+  },
+  {
+   "file": "traces/033.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "7b5ccd4a81403641ad3cb212176bdd7fe86f8703847aa5d7a71a8abec93e992f"
+  },
+  {
+   "file": "traces/034.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "2448deb03c536e0895ff379b29e8ead031179a7c1bd2964deba6a9ea170a7bd9"
+  },
+  {
+   "file": "traces/035.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "77f6d45051bff844f7989490824359b4ebe05d51cf0e5dfdbf432e2edc8c7603"
+  },
+  {
+   "file": "traces/036.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "4994f843f23306a996f04a07246356fe6541f3b71b9374165a1e6744bca2a54b"
+  },
+  {
+   "file": "traces/037.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "153b83b3842dee093866887356079c14379da084ab90dd551729d021932d81f0"
+  },
+  {
+   "file": "traces/038.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "191bf5ba130ea4f8d56eed5e2742fc19b66fea89cdb1b0b0a50fb1cba252884c"
+  },
+  {
+   "file": "traces/039.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "be6587a893efd9358e7e7d6c9aa3a167fab8a9c4ac1ef875fc4380d257d5bc13"
+  },
+  {
+   "file": "traces/040.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "5d31dc253ddbc59600de864b637fe0b4dc07b5aed86cdeb54f5ca1e980b9a675"
+  },
+  {
+   "file": "traces/041.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "6ae0c21eb8cba93f31ec20e2637aea3dcf3e45aacf1a8072eace0759daf35f80"
+  },
+  {
+   "file": "traces/042.parquet",
+   "rows": 128,
+   "unique_rows_added": 128,
+   "sha256": "c62b1a5e40e469b8e4ac098a2eff4483cd6d41012e1563c2c18edec94dbc7158"
+  },
+  {
+   "file": "traces/043.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "ece49f6f54a158758e7b59b235b631956706cab16dd91325f66cab65adbdeb2c"
+  },
+  {
+   "file": "traces/044.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "71f44f79f211f97d68f5ff2a88758c64c7ad66316fff9924efd7aa91a8a8a599"
+  },
+  {
+   "file": "traces/045.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "d85cc0d4c4241ad5c0910b213bdcab2567d69c29e180230e4143dcb92899d669"
+  },
+  {
+   "file": "traces/046.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "2c9ad06dcb7f214c9c315e79f05742b194a9c16094b269e27047869889c454de"
+  },
+  {
+   "file": "traces/047.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "36b48eaba63ea4c1a92a374fc05088441447bc0d9e29b5d5cdd6c697f89a252f"
+  },
+  {
+   "file": "traces/048.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "f926ee8465d42685292a1975e91961a101961aac11ce3e4087ba8deb5ae01aa3"
+  },
+  {
+   "file": "traces/049.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "96eeb89c8275bdf9798a37e57feed7434cde70088f437d36bc6a37dc62e4e62c"
+  },
+  {
+   "file": "traces/050.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "2ba69eaa7bbf986d5bf1a16f365ce5eb48fa7b41a6c6645d04ea0bd479362f84"
+  },
+  {
+   "file": "traces/051.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "310564df6f8919c52241d34319e50b25612c79ab9008c8f6bfc85c29d8837233"
+  },
+  {
+   "file": "traces/052.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "84ac35171ee16ec7d194bf3a16f7dfecfef21b885e592b32f7b171c8a520aec4"
+  },
+  {
+   "file": "traces/053.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "bb1a8191f8cfaaff2fe1107a62f5f0b303ea9f0d16978635ce220fa2943ddd4b"
+  },
+  {
+   "file": "traces/054.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "72320537ffe85b0ef8d1641411f533fb0760f28f33bee06d06734be0c83e2441"
+  },
+  {
+   "file": "traces/055.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "bebdecd805797f3acbd874c2e583353b7d05d4819ed9a689432d1718c1cacf61"
+  },
+  {
+   "file": "traces/056.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "df0a0fee5560ff8c7e2b2337a10a7cf6e2dbce2257ee9938baf3e52ddaeab2cc"
+  },
+  {
+   "file": "traces/057.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "d98ffecbc08c1c88ea9324bc334fbb837c62634a04137b4daf869e36eeef68ba"
+  },
+  {
+   "file": "traces/058.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "3f7b756ec9a81bf51eadd6e8e3c0c1f59d7cf0a478d6a70825e88b68753f9ff7"
+  },
+  {
+   "file": "traces/059.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "0f4bbf8274d25e99c7b8ebd561e598370a01cb504f6a0bd38f523ff50ef70a6a"
+  },
+  {
+   "file": "traces/060.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "ca9ee828a8fbcfaf070cc877d43c5473399134c920eaa8b419e5a1f916e73278"
+  },
+  {
+   "file": "traces/061.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "f112584bbcc16f74c7c2be24315a4ea01466a6d5ac23d45091a0334d98b55802"
+  },
+  {
+   "file": "traces/062.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "7d1499fbebf3fbd6652276a9d7e5774f1fbd89eec8219f9fd004ed63c936269a"
+  },
+  {
+   "file": "traces/063.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "5b242bb75d25a79e3da9fd3661ffa6d9d95dbfcc1aa0c50e6864db8d2f1ed12c"
+  },
+  {
+   "file": "traces/064.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "b602b51792275713ad77d48ac71bed30fd48d90d25be0ba185918637b9140085"
+  },
+  {
+   "file": "traces/065.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "1217c3037268d153ea7436c231f00465652fc755fbf364f772b5021ce65311b5"
+  },
+  {
+   "file": "traces/066.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "f01d50772663013f4f440ca834431faa2985c53f2f2b9824ba50e0f9f58f5a29"
+  },
+  {
+   "file": "traces/067.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "b6fe27dc14ddf19f2a57e7681b4177b158303711d78dfcbde568808c3caff01a"
+  },
+  {
+   "file": "traces/068.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "d97e7468897ae71a20b9b5e0d7b3a8b8abe7cb59d0f6bbc1e69aebf2dfbafcae"
+  },
+  {
+   "file": "traces/069.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "5dfd51752166e59b089f01269051a04214f7238677fae1e872627b52b4328207"
+  },
+  {
+   "file": "traces/070.parquet",
+   "rows": 96,
+   "unique_rows_added": 96,
+   "sha256": "e95f50cd6e9d91cff6631d92c1b86f635da6161b434f36e5cf9574256a78d99e"
+  },
+  {
+   "file": "traces/071.parquet",
+   "rows": 16,
+   "unique_rows_added": 16,
+   "sha256": "1bf864c55d69dfd42cd0b50768359f3545bcc91335033c0985cd5f75179c3eb3"
+  },
+  {
+   "file": "traces/072.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "6d738b011c9062a3281fa3698407216b1f46688808beec3eaa3edea1631f1d56"
+  },
+  {
+   "file": "traces/073.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "2f206e55f4bd7a1546e34fb53be044d3f57f6277707ac6b9464bf5d6ae83bf97"
+  },
+  {
+   "file": "traces/074.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "b74c2790701f774929cae272b7bb587d118c414641e3aaf454487ad8372761f9"
+  },
+  {
+   "file": "traces/075.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "8592a094a818106c31b1665045eb5807ec1c3c289cf11ab237e6f210dfb7e148"
+  },
+  {
+   "file": "traces/076.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "c42ce85e82d378039b2e08a781ee2d2595f049cd135ff36d72459622e89dc7f7"
+  },
+  {
+   "file": "traces/077.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "ae18ef2d242ac31e9faf8e74cd19523ec2f5a506a5be8b05839d93bf5b9c1904"
+  },
+  {
+   "file": "traces/078.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "916c71fa7acceafffb7b9a908f9bb78a26284912f705967fd9e4195c0722707e"
+  },
+  {
+   "file": "traces/079.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "6ba3251b4ad151b9a8a8fd6f87f9657245e402cc7a3e6ceec91e9eaa3db2f08a"
+  },
+  {
+   "file": "traces/080.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "6bf54e4c7cbbfaeb8deb4ebaa04d276e8b697d6eb51b139fab5e39c991f033ca"
+  },
+  {
+   "file": "traces/081.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "69c31836a6328c70ba58a49942afeeabfa501b43114865f721ba4d5d3fda46d3"
+  },
+  {
+   "file": "traces/082.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "f67f6dd275701c623d23e1163ccc18acd395c432321ead5028fa5dc365b3d69a"
+  },
+  {
+   "file": "traces/083.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "061daad0415f74dbb90e21bb5d19a0a23480701e9f192a73fa383f6f0efe408e"
+  },
+  {
+   "file": "traces/084.parquet",
+   "rows": 384,
+   "unique_rows_added": 384,
+   "sha256": "50cc03f60a5ee641a23864c5684c52c9fd46daf53fba75fd47a915104480252a"
+  },
+  {
+   "file": "traces/085.parquet",
+   "rows": 128,
+   "unique_rows_added": 128,
+   "sha256": "a6f2841836816d7f918f38bd0117ce89a8fa4c6070dbb925edb5ceba1b25f8d1"
+  }
+ ],
+ "figures": {
+  "all-models": {
+   "trace_distributions": "plots/all-models-trace-distributions.png",
+   "trace_outcomes": "plots/all-models-trace-outcomes.png",
+   "trace_commands": "plots/all-models-trace-commands.png"
+  },
+  "qwen3-30b-instruct-2507": {
+   "trace_distributions": "plots/qwen3-30b-instruct-2507-trace-distributions.png",
+   "trace_outcomes": "plots/qwen3-30b-instruct-2507-trace-outcomes.png",
+   "trace_commands": "plots/qwen3-30b-instruct-2507-trace-commands.png"
+  },
+  "qwen3-30b-thinking": {
+   "trace_distributions": "plots/qwen3-30b-thinking-trace-distributions.png",
+   "trace_outcomes": "plots/qwen3-30b-thinking-trace-outcomes.png",
+   "trace_commands": "plots/qwen3-30b-thinking-trace-commands.png"
+  }
+ },
+ "metric_implementation_sha256": "ffa2ef63f6c84b193cd50503ece64f6819abc82b894c041ae9c7fb601ef33fe1",
+ "outcome_implementation_sha256": "570e8bdb6930588cb87b9f3728518cae549fd20da595808a2ecef7fed75dddc0",
+ "plotting": {
+  "implementation_sha256": "3e3bdc00f8b2100de4f96bdb48ce6fd87eff518a6bb380f9b07c3da352ad7db6"
+ },
+ "outcome_policy": "exclusive-outcomes-v1"
+}