crosscodeeval_v2: refreshed release with current v1 tasks, uniform card and trace statistics

crosscodeeval_v2/README.md

@@ -2,39 +2,20 @@
 
 ## Dataset stats
 
-- **6,710 v1 tasks → 863 kept; 5,847 newly archived.** All 863 retain their v1 repairs; v2 changes no task instructions, environments, verifiers or reference solutions.
-- **7,763 original tasks: 863 kept and 6,900 archived**, including 1,053 earlier v1 exclusions.
-- **Images:** the original 3 image definitions were not consolidated. Kept tasks reuse the 3 [v1 cached images](https://huggingface.co/datasets/FWeindel/validated-tasks/tree/main/crosscodeeval_v1/cached_images), pinned by immutable revision and checksum in the Parquet. No image copies are needed in v2.
-
-## Dataset characteristics
-
-Cross-file code completion in C#, Java, Python and TypeScript, with related-file context and executable verifiers.
-
-<details><summary>Domains</summary>
-
-Software Engineering; Code Completion; Cross-file Repository Understanding.
-
-</details>
-
-<details><summary>Capabilities</summary>
-
-Read related files, resolve identifiers, follow existing code patterns and complete missing statements.
-
-</details>
-
-**Likely transfer targets:** cross-file code completion and repository understanding.
+- **v1 tasks evaluated:** 6,710
+- **Kept:** 863 (solved at least once by either model)
+- **Excluded:** 5,847 (no recorded success)
+- **Attempts:** pass@16 with Qwen3-30B Instruct and Qwen3-30B Thinking (214,720 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. Recorded terminal failures count as unsuccessful attempts; unsolved tasks are not necessarily defective.
-
 | Model | Tasks evaluated | Attempts | Tasks solved at least once |
 | --- | ---: | ---: | ---: |
 | Qwen3-30B Instruct | 6,710 | 107,360 | 641 |
 | Qwen3-30B Thinking | 6,710 | 107,360 | 627 |
-| Either model | 6,710 | 214,720 | **863** |
+| Either model | 6,710 | 214,720 | 863 |
 
-Successful-attempt distribution **among the 863 kept tasks**. Each model column totals 863; 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 |
 | --- | ---: | ---: |
@@ -56,43 +37,36 @@ Successful-attempt distribution **among the 863 kept tasks**. Each model column
 | 15/16 | 7 | 7 |
 | 16/16 | 13 | 15 |
 
-The kept tasks inherit v1's rule-based repairs, build checks, reference reward 1 and no-op reward 0. Teacher runs used the same task contents without the timeout/anti-cheating sentence; v2 preserves that v1 sentence. [V1 repair details](https://huggingface.co/datasets/FWeindel/validated-tasks/blob/main/crosscodeeval_v1/README.md) · [Teacher results](https://huggingface.co/datasets/FWeindel/validated-tasks/discussions/6).
-
 ## 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)]
-
-# Alternatively, at least 3 successes for either model:
-# filters = [[("instruct_successes", ">=", 3)],
-#            [("thinking_successes", ">=", 3)]]
-
-subset = pq.read_table("tasks.parquet", filters=filters)
-pq.write_table(subset, "selected.parquet")
+subset = pq.read_table("tasks.parquet", filters=[("instruct_successes", "=", 1)])
 ```
 
-`patch.py` reproduces the release from the pinned v1 Parquets and recorded counts.
-
-## 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).
-
 ## Traces
 
-[Trace Parquets](traces) contain **214,720 attempts**, including successful and failed attempts from every evaluated task, kept or excluded. Zstandard-compressed, one row per attempt; **1** 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.
+`traces/` holds all 214,720 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. 1 attempts have no saved conversation (`trajectory_found=false`).
 
 ```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"))
+
+traces = ds.dataset("traces", format="parquet")
+failed = traces.to_table(filter=ds.field("reward") == 0)
 conversation = json.loads(failed["trajectory"][0].as_py())
 ```
+
+Trajectory statistics (plots, command distribution) are in [trace_stats](trace_stats/README.md).
+
+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).
+
+`patch.py` rebuilds `tasks.parquet` from the current v1 tasks and the recorded results, both pinned to revision `a43f5635`.

crosscodeeval_v2/patch.py

@@ -1,140 +1,86 @@
-"""Reproduce CrossCodeEval v2 from pinned v1 Parquets and the recorded pass@16 results.
+"""Reproduce CrossCodeEval 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:
+- crosscodeeval_v1/tasks.parquet: the current v1 payloads, content hashes and image manifest.
+- crosscodeeval_v2/tasks.parquet as published before this refresh: which tasks are kept, the
+  recorded success counts (`instruct_successes`, `thinking_successes`, `attempts_per_model`, `archive_category`) 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
-import io
-import tarfile
 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': ('crosscodeeval_v1/tasks.parquet', '2607baee26489948dbfeb1a02494e5a7d149752a1910be882c40cbc09f139f0f'),
+           'v2': ('crosscodeeval_v2/tasks.parquet', '21a25eac1b374c251f67f76e7abbb865435eeba329c7598ba210ba274d31f648')}
+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) == 6710 and len(kept) == 863
-    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())).append(pa.field('archive_category', pa.string()))
-    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 result['models'][model]['complete'] and result['models'][model]['missing'] == 0
-                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'])
-            # V1's original-source archive predates the validation columns.
-            # Preserve its category and reason while computing its content hash.
-            with tarfile.open(fileobj=io.BytesIO(row['task_binary']), mode='r:*') as archive_tar:
-                files = [(str(Path(m.name)), archive_tar.extractfile(m).read())
-                         for m in archive_tar.getmembers() if m.isfile()]
-            digest = hashlib.sha256()
-            for name, content in sorted(files):
-                encoded = name.encode()
-                digest.update(len(encoded).to_bytes(8, 'big')); digest.update(encoded)
-                digest.update(len(content).to_bytes(8, 'big')); digest.update(content)
-            row['content_sha256'] = digest.hexdigest()
-            row['archive_stage'] = 0
-            row['stages_passed'] = ''
-            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': 863, 'newly_archived': 5847, 'prior_archived': 1053}
-    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': 'crosscodeeval_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)

crosscodeeval_v2/tasks.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21a25eac1b374c251f67f76e7abbb865435eeba329c7598ba210ba274d31f648
-size 11270178
+oid sha256:a8493ea7ff4169db8f76ca0079f6fae17fe12bd752eb45887ac6ea8beb094781
+size 1940558

crosscodeeval_v2/trace_stats/README.md

@@ -0,0 +1,34 @@
+# Trajectory statistics
+
+214,720 attempts across 6,710 tasks of crosscodeeval_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 | 107,360 | 107,360 |
+| Median agent turns | 2 | 2 |
+| Median input tokens per attempt | 3,865 | 3,723 |
+| Median output tokens per attempt | 393 | 1,504 |
+| Median agent runtime (seconds) | 6 | 20 |
+| Median verifier runtime (seconds) | 1 | 1 |
+
+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: 1; without a verifier reward: 2. Their missing measurements stay empty.

crosscodeeval_v2/trace_stats/data/per_attempt.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2885fa1ef1aed0e7ca13cd0b4c481f56b45e9924ef8c340680661f008f10fd82
+size 20413738

crosscodeeval_v2/trace_stats/data/per_task.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a89e0f5513a0ff2c1bc150e36fd9130fa45e38bd841607e32fb472fda529735
+size 3992643

crosscodeeval_v2/trace_stats/data/provenance.json

@@ -0,0 +1,80 @@
+{
+ "deduplication_key": [
+  "model",
+  "task_id",
+  "attempt_id"
+ ],
+ "sources": [
+  {
+   "file": "traces/000.parquet",
+   "rows": 21648,
+   "unique_rows_added": 21648,
+   "sha256": "f5ec9a5f6626dbe3acb070973178d93222dcaba4877f8f2956c0e779f2e57c35"
+  },
+  {
+   "file": "traces/001.parquet",
+   "rows": 30320,
+   "unique_rows_added": 30320,
+   "sha256": "0f128724a06512d443d07a1a14e0fac93147568d6b5483c961b792acf28ed9d7"
+  },
+  {
+   "file": "traces/002.parquet",
+   "rows": 6912,
+   "unique_rows_added": 6912,
+   "sha256": "763667630db2a35822fe309e2d5a4dcbc2063d838f4535ec31c0d4eb34a1e32d"
+  },
+  {
+   "file": "traces/003.parquet",
+   "rows": 48480,
+   "unique_rows_added": 48480,
+   "sha256": "b78e3c72820fa123db88e314d936f0fe2c515c6e83a30267a343c3e93511840b"
+  },
+  {
+   "file": "traces/004.parquet",
+   "rows": 21648,
+   "unique_rows_added": 21648,
+   "sha256": "fba5ef560603b92e4129d36fd67eb679b45e44e3aca97698a2c4f1ccb851a5c7"
+  },
+  {
+   "file": "traces/005.parquet",
+   "rows": 30320,
+   "unique_rows_added": 30320,
+   "sha256": "5bed19d7e31c40122b670f154f73223613115717da5e9580d2e33ff165bbf27d"
+  },
+  {
+   "file": "traces/006.parquet",
+   "rows": 6912,
+   "unique_rows_added": 6912,
+   "sha256": "7ae811ffc97006b2ff52ade98a88911bb6c1d26aa9e5ea6e555ef8ea2b0e329d"
+  },
+  {
+   "file": "traces/007.parquet",
+   "rows": 48480,
+   "unique_rows_added": 48480,
+   "sha256": "851ad461e211c2580e47dd077f1adf0c45676c425bf0eb007fa2bc5c66b15008"
+  }
+ ],
+ "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"
+}

crosscodeeval_v2/trace_stats/data/summary.json

@@ -0,0 +1,781 @@
+{
+ "scope": "All attempts in the published trace Parquets of this folder.",
+ "created_at": "2026-10-17T19:07:21.398219+00:00",
+ "input_rows": 214720,
+ "duplicate_records_removed": 0,
+ "unique_attempts": 214720,
+ "unique_tasks": 6710,
+ "missing_trajectories": 1,
+ "null_verifier_rewards": 2,
+ "all_attempts": {
+  "trajectories": 214720,
+  "graded": 214720,
+  "solved": 5888,
+  "turns": {
+   "n": 214719,
+   "mean": 2.4243127063743777,
+   "median": 2,
+   "p90": 2,
+   "p95": 3,
+   "max": 433
+  },
+  "input_tokens": {
+   "n": 214719,
+   "mean": 8297.892450132498,
+   "median": 3794,
+   "p90": 7302,
+   "p95": 9526,
+   "max": 6431947
+  },
+  "input_tokens_total": 1781715169,
+  "output_tokens": {
+   "n": 214719,
+   "mean": 1419.1730307983923,
+   "median": 872,
+   "p90": 2317,
+   "p95": 3031,
+   "max": 171157
+  },
+  "output_tokens_total": 304723414,
+  "peak_context_tokens": {
+   "n": 214668,
+   "mean": 3159.26866603313,
+   "median": 2577,
+   "p90": 4254,
+   "p95": 5140,
+   "max": 337979
+  },
+  "peak_context_fraction": {
+   "n": 214668,
+   "mean": 0.09641322833353058,
+   "median": 0.078643798828125,
+   "p90": 0.12982177734375,
+   "p95": 0.1568603515625,
+   "max": 10.314300537109375
+  },
+  "trajectories_above_90_percent_context": 1464,
+  "latency_seconds": {
+   "model_call": {
+    "n": 524489,
+    "mean": 6.043952373520901,
+    "median": 3.739053726196289,
+    "p90": 13.835546731948853,
+    "p95": 19.431769132614136,
+    "max": 684.4818048477173
+   },
+   "tool_call": null,
+   "agent": {
+    "n": 214720,
+    "mean": 19.80614744521703,
+    "median": 11.3006,
+    "p90": 30.660667,
+    "p95": 40.151806,
+    "max": 3310.508183
+   },
+   "verifier": {
+    "n": 214718,
+    "mean": 1.292768486615002,
+    "median": 1.233642,
+    "p90": 1.377689,
+    "p95": 1.418958,
+    "max": 118.029042
+   },
+   "trial": {
+    "n": 214720,
+    "mean": 37.41115042656483,
+    "median": 21.542612,
+    "p90": 58.177431,
+    "p95": 112.493342,
+    "max": 3318.148226
+   }
+  },
+  "tools": {
+   "bash_command": {
+    "calls": 296320,
+    "calls_per_trajectory": 1.3800298062593144,
+    "failed": null,
+    "failure_rate": null
+   },
+   "mark_task_complete": {
+    "calls": 434722,
+    "calls_per_trajectory": 2.024599478390462,
+    "failed": null,
+    "failure_rate": null
+   }
+  },
+  "tool_steps": 472933,
+  "tool_steps_without_output": 0,
+  "tool_steps_with_error_text": 9827,
+  "tool_steps_with_error_text_rate": 0.020778841823260376,
+  "terminations": {
+   "harness_error": 1,
+   "session_ended": 34,
+   "task_complete": 214484,
+   "task_timeout": 201
+  },
+  "command_distribution": {
+   "                        DELETE": 3.636079361068135e-07,
+   "                    matcher.group(\"artifactId\"),\n": 2.1485923497220796e-07,
+   "                    matcher.group(\"classifier\"),\n": 2.1485923497220796e-07,
+   "                    matcher.group(\"groupId\"),\n": 2.1485923497220796e-07,
+   "                    matcher.group(\"type\")),\n": 2.1485923497220796e-07,
+   "                    matcher.group(\"version\"),\n": 2.1485923497220796e-07,
+   "                .collect(Collectors.joining(\"\\n\"));": 7.878171948980958e-07,
+   "                .collect(Collectors.joining(\"\\n\"));\n": 7.878171948980958e-07,
+   "                .map(entry -> \"\\u2022 \" + entry.getDisplayValue() + \" (\" + entry.getRank() + \")\")\n": 1.5756343897961916e-06,
+   "                Collectors.toCollection(LinkedHashSet::new)))));\n": 2.1485923497220796e-07,
+   "                matcher -> new Dependency(\n": 2.1485923497220796e-07,
+   "                throw new UnsupportedOperationException(Unsupported": 3.636079361068135e-07,
+   "            Collectors.mapping(\n": 2.1485923497220796e-07,
+   "            ex.printStackTrace();\n": 3.636079361068135e-07,
+   "            matcher -> matcher.group(\"module\"),\n": 2.1485923497220796e-07,
+   "            preparedStatement.close();\n": 3.636079361068135e-07,
+   "            preparedStatement.executeUpdate();\n": 3.636079361068135e-07,
+   "            }\n": 3.636079361068135e-07,
+   "            } else {\n": 3.636079361068135e-07,
+   "        .collect(Collectors.groupingBy(\n": 2.1485923497220796e-07,
+   "        .filter(Matcher::matches)\n": 2.1485923497220796e-07,
+   "        .map(line -> DEPENDENCY_COPY_ITEM_PATTERN.matcher(line))\n": 2.1485923497220796e-07,
+   "        }\n": 3.636079361068135e-07,
+   "        } catch (SQLException ex) {\n": 3.636079361068135e-07,
+   "    if (newVotes.length === 0) break;\n": 5.252114632653972e-07,
+   "    page++;\n": 5.252114632653972e-07,
+   "    resultsSize += newVotes.length;\n": 5.252114632653972e-07,
+   "    votes = votes.concat(newVotes);\n": 5.252114632653972e-07,
+   "    | Value Contains Null: ": 1.1817257923471438e-06,
+   "    | Value Type: ": 1.1817257923471438e-06,
+   "    }\n": 3.636079361068135e-07,
+   "  return votes;\n": 5.252114632653972e-07,
+   "  } while (page < maxPage);\n": 5.252114632653972e-07,
+   " && !key) {\n                        key = true;\n                        if (currentKey != null && currentValue != null) {\n                            cookieStore.add(currentKey.toString(), currentValue.toString());\n                        }\n                        currentKey = null;\n                        currentValue = null;\n                    } else if (key) {\n                        if (currentKey != null) currentKey.append(character);\n                        else currentKey = new StringBuilder().append(character);\n                    } else {\n                        if (currentValue != null) currentValue.append(character);\n                        else currentValue = new StringBuilder().append(character);\n                    }\n                }\n                if (currentKey != null && currentValue != null) {\n                    cookieStore.add(currentKey.toString(), currentValue.toString());\n                }\n            });\n        }\n    });": 7.878171948980958e-07,
+   " + ag.getEndereco() + n;": 1.1817257923471438e-06,
+   " + ag.getNome() + ": 1.1817257923471438e-06,
+   " + conta.getAgencia().getNumAgencia() + ": 7.878171948980958e-07,
+   " + conta.getNumConta() + ": 1.1817257923471438e-06,
+   " + conta.getSaldo() + ": 7.878171948980958e-07,
+   " + conta.getTipoDeConta() + ": 7.878171948980958e-07,
+   " + e.getStackTrace())": 7.878171948980958e-07,
+   " + response.getBody().length() + rn, StandardCharsets.UTF_8);\nwrite(rn, StandardCharsets.UTF_8);\nwrite(response.getBody(), StandardCharsets.UTF_8);\nflush();\nclose();": 6.75271881341225e-07,
+   " + url.toString();\n    }\n}\n": 1.5756343897961916e-06,
+   " 4;\n    }\n\n    public Response respond(String userInput) {\n        List<Map<String, String>> fullPrompt = getFullPrompt(userInput);\n        return openAIModel.generate(fullPrompt);\n    }\n\n    public void addGoal(String goal) {\n        goals.add(goal);\n    }\n\n    public String getName() {\n        return name;\n    }\n\n    public String getDescription() {\n        return description;\n    }\n\n    public List<String> getGoals() {\n        return goals;\n    }\n}\n": 1.5756343897961916e-06,
+   ") : item;\n    if (seen.has(key)) {\n      field.report(messages.array.distinct, \"array.distinct\", field, options);\n      return;\n    }\n    seen.add(key);\n  }\n": 1.5756343897961916e-06,
+   "))\n        })\n    }\n": 2.3634515846942875e-06,
+   ")) }\n      : { success: true, value: result }\n  }": 2.3634515846942875e-06,
+   ");": 4.975687546724815e-06,
+   ")}\n            headers.append(\"Set-Cookie\", cookieHeader)\n        })\n    }": 1.1817257923471438e-06,
+   ")}\n            headers.set(\"Set-Cookie\", cookieHeader)\n        })\n    }\n": 1.1817257923471438e-06,
+   ")}\n            headers.set(Set-Cookie, cookieHeader)\n        })\n    }\n": 1.1817257923471438e-06,
+   ")}\n  }": 2.3634515846942875e-06,
+   ")}  }": 1.5756343897961916e-06,
+   ")}  };\n": 9.453806338777151e-07,
+   ")}  };\n}\n": 1.5756343897961916e-06,
+   "7 + event.getMessage()));": 1.3610499108789354e-06,
+   "8\u2192": 1.3610499108789354e-06,
+   "9Particle": 2.3634515846942875e-06,
+   ":wq": 1.6731736615454797e-06,
+   "=": 2.626057316326986e-07,
+   "C-o": 6.5501372490098824e-06,
+   "C-x": 7.495517882887597e-06,
+   "CPF:": 3.243180785663828e-06,
+   "Conta:": 1.6675463958676361e-06,
+   "Dias": 7.878171948980958e-07,
+   "EOF": 8.36298253045671e-07,
+   "EOL": 1.0742961748610398e-07,
+   "ESC": 6.75271881341225e-07,
+   "Exception": 3.1512687795923832e-06,
+   "Merge": 1.5756343897961916e-06,
+   "Saldo:": 1.1817257923471438e-06,
+   "Skip": 4.332994571939527e-06,
+   "String": 4.9020777849977344e-06,
+   "Taxa": 7.878171948980958e-07,
+   "Valor": 1.5756343897961916e-06,
+   "[assignment]": 7.55096352937472e-07,
+   "[no_commands]": 0.0021129257167166932,
+   "[unknown]": 0.004445344469903186,
+   "^O": 2.8417691673109883e-06,
+   "^X": 3.787149801188703e-06,
+   "a + player.getName() + ": 1.3610499108789354e-06,
+   "adb": 2.3634515846942875e-06,
+   "apt": 2.757360182143335e-06,
+   "awk": 2.297456427006489e-06,
+   "base64": 8.440898516765312e-08,
+   "bash": 8.679791871013388e-08,
+   "break": 3.849491077815793e-06,
+   "cPlayer": 5.953647087158467e-06,
+   "cat": 0.024815864726879272,
+   "cd": 0.005623943229506036,
+   "chmod": 7.336287105929358e-07,
+   "cp": 1.0165537392499647e-05,
+   "curl": 2.675553684592835e-06,
+   "currentClass": 1.5756343897961917e-07,
+   "dd": 1.5263141463686016e-06,
+   "deleted": 1.1817257923471438e-06,
+   "double": 4.609444164169885e-05,
+   "e + distance + m": 6.805249554394677e-07,
+   "echo": 0.24787408132331754,
+   "ed": 1.0832486429848817e-06,
+   "exit": 0.00037404716323909907,
+   "fetch": 2.626057316326986e-07,
+   "find": 0.0008687807532575533,
+   "grep": 0.00026059047777050274,
+   "h": 5.252114632653972e-07,
+   "hasDuplicateHue": 1.3130286581634932e-06,
+   "head": 7.414905803744433e-05,
+   "hexdump": 4.459342612630731e-08,
+   "huffmanMetaCodes": 4.7269031693885753e-07,
+   "huffmanXSize": 9.979017802042547e-07,
+   "huffmanYSize": 9.979017802042547e-07,
+   "i": 1.1479621982800826e-06,
+   "int": 4.323319623932007e-06,
+   "java": 7.878171948980958e-07,
+   "linha": 1.0988271387769339e-05,
+   "ls": 0.009346892369496068,
+   "mkdir": 0.0001938821400078831,
+   "mv": 6.938324672566562e-05,
+   "n": 2.238989708525513e-05,
+   "nano": 0.00010135158204102967,
+   "newQuery": 2.3634515846942875e-06,
+   "nextPrefix": 1.1817257923471438e-06,
+   "nodo1": 1.1817257923471438e-06,
+   "nodo2": 1.1817257923471438e-06,
+   "nodo3": 1.1817257923471438e-06,
+   "owner": 3.70274081602105e-06,
+   "printf": 0.0006508597001769322,
+   "pwd": 3.560388046691845e-05,
+   "python": 5.337205303524393e-06,
+   "r)": 1.5756343897961916e-06,
+   "rT": 2.3634515846942875e-06,
+   "readlink": 8.516942647546982e-08,
+   "return": 3.914261596348163e-05,
+   "rm": 9.979017802042547e-07,
+   "sed": 5.614607807661492e-05,
+   "set": 1.6873781392314991e-06,
+   "setup": 9.453806338777151e-07,
+   "sleep": 4.7269031693885753e-07,
+   "solution.txt": 6.75271881341225e-07,
+   "sort": 6.583193673325983e-06,
+   "strace": 4.258471323773491e-08,
+   "student": 8.032645908764899e-07,
+   "tail": 0.00012615629733020718,
+   "task_complete": 0.0018277358921635824,
+   "task_complete:": 0.00024028424444391924,
+   "task_completen": 7.090354754082863e-06,
+   "tee": 0.0001006543088358318,
+   "this.leaderboardversion": 7.878171948980958e-07,
+   "throw": 3.352413595311046e-08,
+   "touch": 2.5255204712848962e-05,
+   "tr": 3.7317656600436116e-07,
+   "true": 0.0002292548037153459,
+   "updated": 1.1817257923471438e-06,
+   "var": 2.7805312761109265e-07,
+   "vi": 1.4180709508165727e-06,
+   "vim": 7.383407195541618e-06,
+   "wget": 3.121020998985475e-07,
+   "which": 7.649265130200632e-06,
+   "write_solution": 0.7000092339178839,
+   "write_solution.sh": 7.336287105929358e-07,
+   "xargs": 0.00020422834240321232,
+   "xxd": 2.4878437733624077e-07,
+   "y": 4.726903169388575e-06,
+   "yyyy": 7.878171948980958e-07,
+   "\u00daltimo": 7.878171948980958e-07
+  },
+  "output_cap_events": 0,
+  "oom_recoveries": 0,
+  "trajectories_with_oom_recovery": 0,
+  "malformed_tool_calls": 34600,
+  "trajectories_with_malformed_tool_calls": 3066,
+  "errors": {
+   "model_server": {
+    "trajectories": 1,
+    "types": {
+     "ContextManagementInfrastructureError": 1
+    }
+   },
+   "environment": {
+    "trajectories": 1,
+    "types": {
+     "TmuxSessionEndedError": 1
+    }
+   },
+   "verifier": {
+    "trajectories": 0,
+    "types": {}
+   },
+   "other": {
+    "trajectories": 0,
+    "types": {}
+   }
+  }
+ },
+ "models": {
+  "qwen3-30b-instruct-2507": {
+   "trajectories": 107360,
+   "graded": 107360,
+   "solved": 3003,
+   "turns": {
+    "n": 107359,
+    "mean": 2.6162874095325033,
+    "median": 2,
+    "p90": 2,
+    "p95": 4,
+    "max": 433
+   },
+   "input_tokens": {
+    "n": 107359,
+    "mean": 11270.863020333647,
+    "median": 3865,
+    "p90": 7488,
+    "p95": 9877,
+    "max": 6431947
+   },
+   "input_tokens_total": 1210028583,
+   "output_tokens": {
+    "n": 107359,
+    "mean": 786.0006427034529,
+    "median": 393,
+    "p90": 616,
+    "p95": 890,
+    "max": 134532
+   },
+   "output_tokens_total": 84384243,
+   "peak_context_tokens": {
+    "n": 107338,
+    "mean": 2840.7634481730606,
+    "median": 2247,
+    "p90": 3575,
+    "p95": 4483,
+    "max": 72569
+   },
+   "peak_context_fraction": {
+    "n": 107338,
+    "mean": 0.08669322046426577,
+    "median": 0.068572998046875,
+    "p90": 0.109100341796875,
+    "p95": 0.136810302734375,
+    "max": 2.214630126953125
+   },
+   "trajectories_above_90_percent_context": 713,
+   "latency_seconds": {
+    "model_call": {
+     "n": 284844,
+     "mean": 2.6475805339975143,
+     "median": 2.2995083332061768,
+     "p90": 4.34476900100708,
+     "p95": 5.078218698501587,
+     "max": 149.47315096855164
+    },
+    "tool_call": null,
+    "agent": {
+     "n": 107360,
+     "mean": 11.691557514307005,
+     "median": 6.294525,
+     "p90": 9.719629,
+     "p95": 13.298929,
+     "max": 2464.816905
+    },
+    "verifier": {
+     "n": 107360,
+     "mean": 1.288768823733234,
+     "median": 1.236909,
+     "p90": 1.375125,
+     "p95": 1.411272,
+     "max": 117.897129
+    },
+    "trial": {
+     "n": 107360,
+     "mean": 28.064082660068927,
+     "median": 13.718699,
+     "p90": 34.028572,
+     "p95": 89.080016,
+     "max": 2501.000071
+    }
+   },
+   "tools": {
+    "bash_command": {
+     "calls": 167067,
+     "calls_per_trajectory": 1.5561382265275707,
+     "failed": null,
+     "failure_rate": null
+    },
+    "mark_task_complete": {
+     "calls": 215763,
+     "calls_per_trajectory": 2.0097149776453054,
+     "failed": null,
+     "failure_rate": null
+    }
+   },
+   "tool_steps": 252089,
+   "tool_steps_without_output": 0,
+   "tool_steps_with_error_text": 7146,
+   "tool_steps_with_error_text_rate": 0.028347131370270024,
+   "terminations": {
+    "session_ended": 4,
+    "task_complete": 107272,
+    "task_timeout": 84
+   },
+   "command_distribution": {
+    "                        DELETE": 7.291978094897803e-07,
+    "                    matcher.group(\"artifactId\"),\n": 4.3088961469850656e-07,
+    "                    matcher.group(\"classifier\"),\n": 4.3088961469850656e-07,
+    "                    matcher.group(\"groupId\"),\n": 4.3088961469850656e-07,
+    "                    matcher.group(\"type\")),\n": 4.3088961469850656e-07,
+    "                    matcher.group(\"version\"),\n": 4.3088961469850656e-07,
+    "                .collect(Collectors.joining(\"\\n\"));": 1.5799285872278573e-06,
+    "                .collect(Collectors.joining(\"\\n\"));\n": 1.5799285872278573e-06,
+    "                .map(entry -> \"\\u2022 \" + entry.getDisplayValue() + \" (\" + entry.getRank() + \")\")\n": 3.1598571744557146e-06,
+    "                Collectors.toCollection(LinkedHashSet::new)))));\n": 4.3088961469850656e-07,
+    "                matcher -> new Dependency(\n": 4.3088961469850656e-07,
+    "                throw new UnsupportedOperationException(Unsupported": 7.291978094897803e-07,
+    "            Collectors.mapping(\n": 4.3088961469850656e-07,
+    "            ex.printStackTrace();\n": 7.291978094897803e-07,
+    "            matcher -> matcher.group(\"module\"),\n": 4.3088961469850656e-07,
+    "            preparedStatement.close();\n": 7.291978094897803e-07,
+    "            preparedStatement.executeUpdate();\n": 7.291978094897803e-07,
+    "            }\n": 7.291978094897803e-07,
+    "            } else {\n": 7.291978094897803e-07,
+    "        .collect(Collectors.groupingBy(\n": 4.3088961469850656e-07,
+    "        .filter(Matcher::matches)\n": 4.3088961469850656e-07,
+    "        .map(line -> DEPENDENCY_COPY_ITEM_PATTERN.matcher(line))\n": 4.3088961469850656e-07,
+    "        }\n": 7.291978094897803e-07,
+    "        } catch (SQLException ex) {\n": 7.291978094897803e-07,
+    "    if (newVotes.length === 0) break;\n": 1.0532857248185714e-06,
+    "    page++;\n": 1.0532857248185714e-06,
+    "    resultsSize += newVotes.length;\n": 1.0532857248185714e-06,
+    "    votes = votes.concat(newVotes);\n": 1.0532857248185714e-06,
+    "    }\n": 7.291978094897803e-07,
+    "  return votes;\n": 1.0532857248185714e-06,
+    "  } while (page < maxPage);\n": 1.0532857248185714e-06,
+    " && !key) {\n                        key = true;\n                        if (currentKey != null && currentValue != null) {\n                            cookieStore.add(currentKey.toString(), currentValue.toString());\n                        }\n                        currentKey = null;\n                        currentValue = null;\n                    } else if (key) {\n                        if (currentKey != null) currentKey.append(character);\n                        else currentKey = new StringBuilder().append(character);\n                    } else {\n                        if (currentValue != null) currentValue.append(character);\n                        else currentValue = new StringBuilder().append(character);\n                    }\n                }\n                if (currentKey != null && currentValue != null) {\n                    cookieStore.add(currentKey.toString(), currentValue.toString());\n                }\n            });\n        }\n    });": 1.5799285872278573e-06,
+    " + ag.getEndereco() + n;": 2.369892880841786e-06,
+    " + ag.getNome() + ": 2.369892880841786e-06,
+    " + conta.getAgencia().getNumAgencia() + ": 1.5799285872278573e-06,
+    " + conta.getNumConta() + ": 2.369892880841786e-06,
+    " + conta.getSaldo() + ": 1.5799285872278573e-06,
+    " + conta.getTipoDeConta() + ": 1.5799285872278573e-06,
+    " + e.getStackTrace())": 1.5799285872278573e-06,
+    " + response.getBody().length() + rn, StandardCharsets.UTF_8);\nwrite(rn, StandardCharsets.UTF_8);\nwrite(response.getBody(), StandardCharsets.UTF_8);\nflush();\nclose();": 1.3542245033381632e-06,
+    " + url.toString();\n    }\n}\n": 3.1598571744557146e-06,
+    " 4;\n    }\n\n    public Response respond(String userInput) {\n        List<Map<String, String>> fullPrompt = getFullPrompt(userInput);\n        return openAIModel.generate(fullPrompt);\n    }\n\n    public void addGoal(String goal) {\n        goals.add(goal);\n    }\n\n    public String getName() {\n        return name;\n    }\n\n    public String getDescription() {\n        return description;\n    }\n\n    public List<String> getGoals() {\n        return goals;\n    }\n}\n": 3.1598571744557146e-06,
+    ") : item;\n    if (seen.has(key)) {\n      field.report(messages.array.distinct, \"array.distinct\", field, options);\n      return;\n    }\n    seen.add(key);\n  }\n": 3.1598571744557146e-06,
+    "))\n        })\n    }\n": 4.739785761683572e-06,
+    ")) }\n      : { success: true, value: result }\n  }": 4.739785761683572e-06,
+    ");": 4.989248170193233e-07,
+    ")}\n            headers.append(\"Set-Cookie\", cookieHeader)\n        })\n    }": 2.369892880841786e-06,
+    ")}\n            headers.set(\"Set-Cookie\", cookieHeader)\n        })\n    }\n": 2.369892880841786e-06,
+    ")}\n            headers.set(Set-Cookie, cookieHeader)\n        })\n    }\n": 2.369892880841786e-06,
+    ")}\n  }": 4.739785761683572e-06,
+    ")}  }": 3.1598571744557146e-06,
+    ")}  };\n": 1.8959143046734288e-06,
+    ")}  };\n}\n": 3.1598571744557146e-06,
+    "7 + event.getMessage()));": 2.7295185694946742e-06,
+    "8\u2192": 2.7295185694946742e-06,
+    ":wq": 1.3542245033381632e-06,
+    "=": 5.266428624092857e-07,
+    "C-o": 1.5799285872278573e-06,
+    "C-x": 3.4758428919012863e-06,
+    "CPF:": 6.504039350754679e-06,
+    "Conta:": 3.3441821762989643e-06,
+    "Dias": 1.5799285872278573e-06,
+    "EOF": 1.6771549618264947e-06,
+    "EOL": 2.1544480734925328e-07,
+    "ESC": 1.3542245033381632e-06,
+    "Exception": 3.1598571744557146e-06,
+    "Merge": 3.1598571744557146e-06,
+    "Saldo:": 2.369892880841786e-06,
+    "Skip": 8.689607229753214e-06,
+    "String": 4.533467972257381e-06,
+    "Taxa": 1.5799285872278573e-06,
+    "Valor": 3.1598571744557146e-06,
+    "[assignment]": 1.5143085500586491e-06,
+    "[no_commands]": 0.0025974025974025974,
+    "[unknown]": 0.004908990875697404,
+    "a + player.getName() + ": 2.7295185694946742e-06,
+    "adb": 4.739785761683572e-06,
+    "awk": 4.607435722962914e-06,
+    "base64": 1.692780629172704e-07,
+    "bash": 1.7406895149040074e-07,
+    "break": 4.560107608626286e-06,
+    "cPlayer": 1.1939746037764807e-05,
+    "cat": 0.04638586375428574,
+    "cd": 0.007175432611753047,
+    "chmod": 1.4712562505402267e-06,
+    "cp": 2.038648462480105e-05,
+    "curl": 5.365691153133351e-06,
+    "currentClass": 3.1598571744557145e-07,
+    "dd": 3.0609478551048392e-06,
+    "deleted": 2.369892880841786e-06,
+    "double": 4.551625368764433e-05,
+    "e + distance + m": 1.3647592847473371e-06,
+    "echo": 0.2140621709995613,
+    "ed": 2.1724018074383036e-06,
+    "exit": 1.0608091942815612e-05,
+    "fetch": 5.266428624092857e-07,
+    "find": 0.001731850337030823,
+    "grep": 0.0004912695394780917,
+    "h": 1.0532857248185714e-06,
+    "hasDuplicateHue": 1.0532857248185714e-06,
+    "head": 0.000148702284321846,
+    "hexdump": 8.942992003176551e-08,
+    "i": 1.3542245033381632e-06,
+    "int": 5.721004568488241e-06,
+    "java": 1.5799285872278573e-06,
+    "linha": 2.085149069522744e-05,
+    "ls": 0.016192310521127812,
+    "mkdir": 0.0002947435179175777,
+    "mv": 8.953492679604563e-05,
+    "n": 9.479571523367144e-06,
+    "nano": 1.4687367964987622e-05,
+    "newQuery": 4.739785761683572e-06,
+    "nodo1": 2.369892880841786e-06,
+    "nodo2": 2.369892880841786e-06,
+    "nodo3": 2.369892880841786e-06,
+    "owner": 7.425664359970928e-06,
+    "printf": 0.001211261325284522,
+    "pwd": 2.944343971929525e-05,
+    "python": 1.0703502398209337e-05,
+    "r)": 3.1598571744557146e-06,
+    "rT": 4.739785761683572e-06,
+    "readlink": 1.708030905111197e-07,
+    "return": 5.6228812686770775e-05,
+    "rm": 2.0012428771552857e-06,
+    "sed": 5.852513869401461e-05,
+    "set": 3.38395376097374e-06,
+    "setup": 1.8959143046734288e-06,
+    "sleep": 9.479571523367144e-07,
+    "sort": 1.3202270713437088e-05,
+    "strace": 8.540154525555985e-08,
+    "student": 1.610907579134286e-06,
+    "tail": 0.00025213846232612465,
+    "tee": 0.00014118800176096701,
+    "this.leaderboardversion": 1.5799285872278573e-06,
+    "throw": 6.723100371182371e-08,
+    "touch": 3.22419012831557e-05,
+    "tr": 7.48387225528985e-07,
+    "updated": 2.369892880841786e-06,
+    "var": 5.576218543157143e-07,
+    "vim": 3.6422308415492373e-06,
+    "wget": 6.259053914497793e-07,
+    "which": 3.4004784822806465e-06,
+    "write_solution": 0.7033376584757735,
+    "write_solution.sh": 1.4712562505402267e-06,
+    "xargs": 0.00040956988318429787,
+    "xxd": 4.989248170193233e-07,
+    "yyyy": 1.5799285872278573e-06,
+    "\u00daltimo": 1.5799285872278573e-06
+   },
+   "output_cap_events": 0,
+   "oom_recoveries": 0,
+   "trajectories_with_oom_recovery": 0,
+   "malformed_tool_calls": 25144,
+   "trajectories_with_malformed_tool_calls": 1436,
+   "errors": {
+    "model_server": {
+     "trajectories": 0,
+     "types": {}
+    },
+    "environment": {
+     "trajectories": 0,
+     "types": {}
+    },
+    "verifier": {
+     "trajectories": 0,
+     "types": {}
+    },
+    "other": {
+     "trajectories": 0,
+     "types": {}
+    }
+   }
+  },
+  "qwen3-30b-thinking": {
+   "trajectories": 107360,
+   "graded": 107360,
+   "solved": 2885,
+   "turns": {
+    "n": 107360,
+    "mean": 2.2323397913561847,
+    "median": 2,
+    "p90": 2,
+    "p95": 3,
+    "max": 158
+   },
+   "input_tokens": {
+    "n": 107360,
+    "mean": 5324.949571535022,
+    "median": 3723,
+    "p90": 7102,
+    "p95": 9203,
+    "max": 2091916
+   },
+   "input_tokens_total": 571686586,
+   "output_tokens": {
+    "n": 107360,
+    "mean": 2052.3395212369596,
+    "median": 1504,
+    "p90": 2906,
+    "p95": 3648,
+    "max": 171157
+   },
+   "output_tokens_total": 220339171,
+   "peak_context_tokens": {
+    "n": 107330,
+    "mean": 3477.797624149818,
+    "median": 2917,
+    "p90": 4658,
+    "p95": 5515,
+    "max": 337979
+   },
+   "peak_context_fraction": {
+    "n": 107330,
+    "mean": 0.10613396069793146,
+    "median": 0.089019775390625,
+    "p90": 0.14215087890625,
+    "p95": 0.168304443359375,
+    "max": 10.314300537109375
+   },
+   "trajectories_above_90_percent_context": 751,
+   "latency_seconds": {
+    "model_call": {
+     "n": 239645,
+     "mean": 10.080907620895976,
+     "median": 7.226458311080933,
+     "p90": 20.114930152893066,
+     "p95": 26.00395703315735,
+     "max": 684.4818048477173
+    },
+    "tool_call": null,
+    "agent": {
+     "n": 107360,
+     "mean": 27.92073737612705,
+     "median": 19.771433,
+     "p90": 38.249375,
+     "p95": 48.429497,
+     "max": 3310.508183
+    },
+    "verifier": {
+     "n": 107358,
+     "mean": 1.2967682240075262,
+     "median": 1.230451,
+     "p90": 1.379742,
+     "p95": 1.430097,
+     "max": 118.029042
+    },
+    "trial": {
+     "n": 107360,
+     "mean": 46.758218193060735,
+     "median": 29.411122,
+     "p90": 70.003692,
+     "p95": 130.925438,
+     "max": 3318.148226
+    }
+   },
+   "tools": {
+    "bash_command": {
+     "calls": 129253,
+     "calls_per_trajectory": 1.203921385991058,
+     "failed": null,
+     "failure_rate": null
+    },
+    "mark_task_complete": {
+     "calls": 218959,
+     "calls_per_trajectory": 2.0394839791356185,
+     "failed": null,
+     "failure_rate": null
+    }
+   },
+   "tool_steps": 220844,
+   "tool_steps_without_output": 0,
+   "tool_steps_with_error_text": 2681,
+   "tool_steps_with_error_text_rate": 0.012139790983680789,
+   "terminations": {
+    "harness_error": 1,
+    "session_ended": 30,
+    "task_complete": 107212,
+    "task_timeout": 117
+   },
+   "command_distribution": {
+    "    | Value Contains Null: ": 2.357045208127092e-06,
+    "    | Value Type: ": 2.357045208127092e-06,
+    ");": 9.428180832508368e-06,
+    "9Particle": 4.714090416254184e-06,
+    ":wq": 1.990393731307322e-06,
+    "C-o": 1.1493401395819723e-05,
+    "C-x": 1.1493401395819723e-05,
+    "Exception": 3.1427269441694556e-06,
+    "String": 5.268689288754676e-06,
+    "[no_commands]": 0.0016310752840239476,
+    "[unknown]": 0.0039842115858487615,
+    "^O": 5.668132524305625e-06,
+    "^X": 7.553768690807299e-06,
+    "apt": 5.499772152296547e-06,
+    "break": 3.1427269441694556e-06,
+    "cat": 0.0033628010640209824,
+    "cd": 0.004080864787670903,
+    "double": 4.6669495120916424e-05,
+    "echo": 0.2815026903842975,
+    "exit": 0.0007355159571960589,
+    "find": 1.0390045745723867e-05,
+    "grep": 3.116197431004389e-05,
+    "hasDuplicateHue": 1.5713634720847278e-06,
+    "huffmanMetaCodes": 9.428180832508368e-07,
+    "huffmanXSize": 1.990393731307322e-06,
+    "huffmanYSize": 1.990393731307322e-06,
+    "i": 9.428180832508368e-07,
+    "int": 2.9332118145581587e-06,
+    "linha": 1.178522604063546e-06,
+    "ls": 0.0025385846259838667,
+    "mkdir": 9.35675522014088e-05,
+    "mv": 4.9340813023460464e-05,
+    "n": 3.523023313884079e-05,
+    "nano": 0.0001875459717349124,
+    "nextPrefix": 2.357045208127092e-06,
+    "printf": 9.349612658904132e-05,
+    "pwd": 4.173092420865014e-05,
+    "return": 2.214904699784948e-05,
+    "sed": 5.377991483209982e-05,
+    "solution.txt": 1.3468829760726238e-06,
+    "tail": 8.571073484098516e-07,
+    "task_complete": 0.003645563255236569,
+    "task_complete:": 0.000479265858985842,
+    "task_completen": 1.414227124876255e-05,
+    "tee": 6.034035732805354e-05,
+    "touch": 1.830638444978708e-05,
+    "true": 0.0004572667703766558,
+    "vi": 2.8284542497525105e-06,
+    "vim": 1.1104301869398743e-05,
+    "which": 1.18750182390403e-05,
+    "write_solution": 0.6966988534284533,
+    "y": 9.428180832508368e-06
+   },
+   "output_cap_events": 0,
+   "oom_recoveries": 0,
+   "trajectories_with_oom_recovery": 0,
+   "malformed_tool_calls": 9456,
+   "trajectories_with_malformed_tool_calls": 1630,
+   "errors": {
+    "model_server": {
+     "trajectories": 1,
+     "types": {
+      "ContextManagementInfrastructureError": 1
+     }
+    },
+    "environment": {
+     "trajectories": 1,
+     "types": {
+      "TmuxSessionEndedError": 1
+     }
+    },
+    "verifier": {
+     "trajectories": 0,
+     "types": {}
+    },
+    "other": {
+     "trajectories": 0,
+     "types": {}
+    }
+   }
+  }
+ },
+ "outcome_policy": "exclusive-outcomes-v1"
+}