termigen_v2: refreshed release with current v1 tasks, uniform card and trace statistics

termigen_v2/README.md

@@ -2,64 +2,63 @@
 
 ## Dataset stats
 
-- **Tasks:** 3,566 original; 3,336 in v1; **2,277 kept in v2**, all previously patched. 1,059 v1 tasks have no recorded successful attempt; 230 earlier exclusions remain in v1.
-- **Images:** reuse v1 environments and its cached images as they become available. No additional images or task repairs in v2.
+- **v1 tasks evaluated:** 3,336
+- **Kept:** 2,277 (solved at least once by either model)
+- **Excluded:** 1,059 (no recorded success)
+- **Attempts:** pass@4 with Qwen3-Coder-30B Instruct and Qwen3-30B Instruct (26,688 in total)
 
-## Dataset characteristics
+## Pass@4 filtering
 
-[TermiGen](https://github.com/ucsb-mlsec/terminal-bench-env/tree/03bddac74ada2fa344df0e93b4a3ace2034294d1) supplies terminal tasks with prepared files and executable verifiers.
+| Model | Tasks evaluated | Attempts | Tasks solved at least once |
+| --- | ---: | ---: | ---: |
+| Qwen3-Coder-30B Instruct | 3,336 | 13,344 | 2,123 |
+| Qwen3-30B Instruct | 3,336 | 13,344 | 1,850 |
+| Either model | 3,336 | 26,688 | 2,277 |
 
-<details><summary>Domains</summary>
+Every evaluated task has 4 attempts per model; 142 Coder and 179 Instruct attempts have no recorded reward and count as unknown, not as failures.
 
-Software Engineering, Debugging, System Administration, Security, Data Processing, Scientific Computing and Machine Learning: repair code, diagnose services, inspect artifacts and configure computational workflows.
-
-</details>
-<details><summary>Capabilities</summary>
-
-Exploration, Error Recovery and Specification Adherence; specialized tasks also involve Reverse Engineering and Adversarial Reasoning. Labels follow [CLI-Universe](https://arxiv.org/abs/2606.22883v1).
-
-</details>
+### Successful attempts among kept tasks
 
-**Intended benchmark transfer:** terminal problem solving for Terminal-Bench; debugging and environment diagnosis for repository-based software engineering.
+| Successful attempts | Coder tasks | Instruct tasks |
+| --- | ---: | ---: |
+| 0/4 | 154 | 427 |
+| 1/4 | 302 | 417 |
+| 2/4 | 338 | 359 |
+| 3/4 | 444 | 421 |
+| 4/4 | 1,039 | 653 |
 
-## Selection
+## Select tasks by successful-attempt count
 
-Keep tasks with at least one recorded reward of 1 from either model, with four attempts per task and model.
+You can select tasks by how often they were solved, for example all tasks Instruct solved exactly once:
 
-| Model | Tasks solved at least once |
-| --- | ---: |
-| Qwen3-Coder-30B-A3B-Instruct | 2,123 / 3,336 |
-| Qwen3-30B-A3B-Instruct-2507 | 1,850 / 3,336 |
-| Either model | 2,277 / 3,336 |
+```python
+import pyarrow.parquet as pq
 
-All 3,336 evaluated task hashes match v1. V2 preserves task payloads exactly. Build, no-op and rule-based checks carry over from [v1](../termigen_v1/README.md); no reference solutions are provided.
+subset = pq.read_table("tasks.parquet", filters=[("instruct_successes", "=", 1)])
+```
 
-Missing rewards remain unknown, not failures (142 Coder attempts and 179 Instruct attempts). Excluded tasks are those without a recorded success, not confirmed defective tasks.
+## Traces
 
-### Successful attempts among kept tasks
+`traces/` holds all 26,688 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. 64 attempts have no saved conversation (`trajectory_found=false`).
 
-| Instruct successes out of 4 | Tasks |
-| --- | ---: |
-| 0 | 427 |
-| 1 | 417 |
-| 2 | 359 |
-| 3 | 421 |
-| 4 | 653 |
-
-Per-task Coder evidence establishes whether a task was solved; exact per-task success counts are not available in the published report, so they are not inferred.
+```python
+import json
+import pyarrow.dataset as ds
 
-## Files and selecting tasks
+traces = ds.dataset("traces", format="parquet")
+failed = traces.to_table(filter=ds.field("reward") == 0)
+conversation = json.loads(failed["trajectory"][0].as_py())
+```
 
-`tasks.parquet` contains kept tasks, `coder_solved`, `instruct_successes`, `instruct_missing_rewards` and `attempts_per_model`. `metadata/selection.json` records every v1 task's selection and pins the source revision. Excluded payloads remain in that v1 revision; no duplicate archive Parquet is uploaded. `patch.py` reproduces this selection.
+Trajectory statistics (plots, command distribution) are in [trace_stats](trace_stats/README.md).
 
-```python
-import pyarrow.parquet as pq
-subset = pq.read_table("tasks.parquet", filters=[("instruct_successes", "=", 1)])
-pq.write_table(subset, "selected.parquet", compression="zstd")
-```
+Teacher results and trajectories come from [PR #20](https://huggingface.co/datasets/FWeindel/validated-tasks/discussions/20); [recorded model results](https://huggingface.co/datasets/FWeindel/validated-tasks/discussions/11).
 
-Saved teacher trajectories from [PR #20](https://huggingface.co/datasets/FWeindel/validated-tasks/discussions/20) are available in [traces/](traces/), with coverage and provenance in [traces/manifest.json](traces/manifest.json). The 70 Parquet shards contain 26,688 attempt records from both models across all 3,336 evaluated v1 tasks. Filter `retained_v2 == True` for the 2,277 tasks kept in v2. The manifest records 64 missing trajectories and 321 ungraded attempts; missing raw verifier rewards remain null. It also documents missing-reward metadata discrepancies for 103 tasks.
+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
 
-Apptainer 1.5.4, Python 3.12.7 / 3.12.14, Harbor 0.8.1; [Marin Harbor bridge](https://github.com/marin-community/harbor/tree/7faf878c14b6d72737579ec936ebf5f74ba3194d) and [runtime patches](https://github.com/MLI-lab/OT-next-data/tree/main/harbor_patches). Teacher runs used Terminus 2; [recorded model results](https://huggingface.co/datasets/FWeindel/validated-tasks/discussions/11).
+- **Versions:** Apptainer **1.5.4**, Python **3.12.7 / 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`.

termigen_v2/patch.py

@@ -1,22 +1,86 @@
-"""Reproduce TermiGen v2. Requires pyarrow and huggingface_hub."""
-import argparse,hashlib,json
+"""Reproduce TermiGen 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
+
+Both inputs are pinned to repository revision a43f5635 and checked by SHA-256:
+- termigen_v1/tasks.parquet: the current v1 payloads, content hashes and image manifest.
+- termigen_v2/tasks.parquet as published before this refresh: which tasks are kept, the
+  recorded success counts (`coder_solved`, `instruct_successes`, `instruct_missing_rewards`, `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
-from huggingface_hub import hf_hub_download
-p=argparse.ArgumentParser(description=__doc__)
-p.add_argument('--out',type=Path,required=True)
-p.add_argument('--source',type=Path)
-a=p.parse_args()
-m=json.loads((Path(__file__).parent/'metadata/selection.json').read_text())
-a.out.mkdir(parents=True,exist_ok=True)
-f=a.source or Path(hf_hub_download(m['repo_id'],m['source_folder']+'/tasks.parquet',revision=m['source_revision'],repo_type='dataset',cache_dir=a.out/'.source-cache'))
-assert hashlib.sha256(f.read_bytes()).hexdigest()==m['source_sha256']
-t=pq.read_table(f);rows=[]
-for r in t.to_pylist():
- s=m['tasks'][r['path']];assert r['content_sha256']==s['content_sha256']
- if s['kept']:rows.append(dict(r,coder_solved=s['coder_solved'],instruct_successes=s['instruct_successes'],instruct_missing_rewards=s['instruct_missing_rewards'],attempts_per_model=4))
-assert len(rows)==2277
-schema=t.schema.append(pa.field('coder_solved',pa.bool_())).append(pa.field('instruct_successes',pa.int64())).append(pa.field('instruct_missing_rewards',pa.int64())).append(pa.field('attempts_per_model',pa.int64()))
-out=pa.Table.from_pylist(rows,schema=schema).replace_schema_metadata({b'ot.release.v2':json.dumps({k:v for k,v in m.items() if k!='tasks'}).encode()})
-pq.write_table(out,a.out/'tasks.parquet',compression='zstd')
+
+REPO = 'FWeindel/validated-tasks'
+REVISION = 'a43f56353751d6d73a93ff448de6981040c4d0c5'
+SOURCES = {'v1': ('termigen_v1/tasks.parquet', '8d8a935d6409bf72fb567d454ce544a1c571365eee5023a44f855279cba1d777'),
+           'v2': ('termigen_v2/tasks.parquet', 'c80fa56b1e2dc7e5f293c4fda90cb265ccb231f9272be214d48089cea82d9019')}
+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 restrict_images(manifest, kept):
+    result = copy.deepcopy(manifest)
+    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):
+    out.mkdir(parents=True, exist_ok=True)
+    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': 'termigen_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__, formatter_class=argparse.RawDescriptionHelpFormatter)
+    parser.add_argument('--source-dir', type=Path)
+    parser.add_argument('--out', type=Path, required=True)
+    args = parser.parse_args()
+    build(args.source_dir, args.out)

termigen_v2/tasks.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c80fa56b1e2dc7e5f293c4fda90cb265ccb231f9272be214d48089cea82d9019
-size 111908336
+oid sha256:f039c872cc13c12207e45fbe3fca6bf0ac0d093b45f89848fc21e11e7d86cc85
+size 122932790

termigen_v2/trace_stats/README.md

@@ -0,0 +1,34 @@
+# Trajectory statistics
+
+26,688 attempts across 3,336 tasks of termigen_v2, kept and excluded, each counted once. Task success counts are in the [dataset overview](../README.md).
+
+| Metric | qwen3-30b-a3b-instruct-2507 | qwen3-coder-30b-a3b-instruct |
+| --- | ---: | ---: |
+| Recorded attempts | 13,344 | 13,344 |
+| Median agent turns | 6 | 21 |
+| Median input tokens per attempt | 20,686 | 132,895 |
+| Median output tokens per attempt | 2,352 | 4,356 |
+| Median agent runtime (seconds) | 53 | 71 |
+| Median verifier runtime (seconds) | 2 | 2 |
+
+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-a3b-instruct-2507](plots/qwen3-30b-a3b-instruct-2507-trace-commands.png), [qwen3-30b-a3b-instruct-2507](plots/qwen3-30b-a3b-instruct-2507-trace-distributions.png), [qwen3-30b-a3b-instruct-2507](plots/qwen3-30b-a3b-instruct-2507-trace-outcomes.png), [qwen3-coder-30b-a3b-instruct](plots/qwen3-coder-30b-a3b-instruct-trace-commands.png), [qwen3-coder-30b-a3b-instruct](plots/qwen3-coder-30b-a3b-instruct-trace-distributions.png), [qwen3-coder-30b-a3b-instruct](plots/qwen3-coder-30b-a3b-instruct-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: 64; without a verifier reward: 321. Their missing measurements stay empty.

termigen_v2/trace_stats/data/per_attempt.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:638e4b55d59565b437186c6cf3aadc2df636fbe406ad9e83064e9a39cb16ea30
+size 4092049

termigen_v2/trace_stats/data/per_task.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9c0ad8e1c2590f047cc003a6ec94b0803156e2ac420f16ac2a546c7bd50670e
+size 2785099

termigen_v2/trace_stats/data/provenance.json

@@ -0,0 +1,452 @@
+{
+ "deduplication_key": [
+  "model",
+  "task_id",
+  "attempt_id"
+ ],
+ "sources": [
+  {
+   "file": "traces/coder-001.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "62980afd0db621409bcf726010733f6f9b2de9c65eaa19c0765b65fd5c057fd0"
+  },
+  {
+   "file": "traces/coder-002.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "0d315e6520eb72d00917ffcad952255c09168ead4303e6b07163cf72f64a6229"
+  },
+  {
+   "file": "traces/coder-003.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "f07b673bd9969cbf6c54db18846277a5ee27761c8206c0409f051455dd26aac8"
+  },
+  {
+   "file": "traces/coder-004.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "f7ea63e25da815ef7ceeb210dc748227e45ef8a14b5573460f7db56ae2926708"
+  },
+  {
+   "file": "traces/coder-005.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "e6939cf7aa2ba3da4ba5b99cee1f6df961aa27d921328e9bfa2798efc321c864"
+  },
+  {
+   "file": "traces/coder-006.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "9f8b95649fbbb42e8624d41b5601f59c3f48eb2d0862cb4fa6bf69c900923f7d"
+  },
+  {
+   "file": "traces/coder-007.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "35b51aed2605c8fa6e3700d35bc03e57c4bc7f1c0ab3d8d2e6550a5d42586ce7"
+  },
+  {
+   "file": "traces/coder-008.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "94c73d1f0b5575c6db0c1ca16030c02d12137b6b9e72963cf84246781eb98317"
+  },
+  {
+   "file": "traces/coder-009.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "ef4144bee0926389330a62ba7533dbdfdd627fe238183a2936ec37b1dad5215f"
+  },
+  {
+   "file": "traces/coder-010.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "be384bea323d5103ca4144ea0cd0aaf3b69b641cbba5d48d8a4465c3cad365f0"
+  },
+  {
+   "file": "traces/coder-011.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "25fa9f053eb591083fec052f9cb7cef2a9b6ab886838f984550fccdb7baf6b62"
+  },
+  {
+   "file": "traces/coder-012.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "a8faab8c456d583eb21f32794e09b3bad82d16f268fc3ae9e8679a7668fea01e"
+  },
+  {
+   "file": "traces/coder-013.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "540d9eebd029cb0e6b70c0dde6125182ad43b51c97093331cb32044cf461a720"
+  },
+  {
+   "file": "traces/coder-014.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "05ee7152d8da7579cf7ba0398cb74de2e40b9f610b49cd2cb2b83d97622b7ee5"
+  },
+  {
+   "file": "traces/coder-015.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "042c9e4e09ffb21cb3282d78d09a22772a7b98462bbba521fb7817e38ec54918"
+  },
+  {
+   "file": "traces/coder-016.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "f9bff34a4b0178c51d2282f0767d4553f4055ff801daf1e3f4a530b613780baa"
+  },
+  {
+   "file": "traces/coder-017.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "e46fd091eb737b24510ca542b4f2a6403cbec0a3d95023d68244e670d2e5450f"
+  },
+  {
+   "file": "traces/coder-018.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "568ece5b039742c8adcd9f7ed3641b2b152f412668218450a3658a71b34e4f27"
+  },
+  {
+   "file": "traces/coder-019.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "c7e4021d7bb02ef58014780de96089d9f03bac09c8aa14290b56b047401c3578"
+  },
+  {
+   "file": "traces/coder-020.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "1b2c27e3e439a76d09904be8fcbb9c9812a9bab1a814ecf8f49c98408835be02"
+  },
+  {
+   "file": "traces/coder-021.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "b852a375e3ca64f09ea311361834655309b4a07ec5e5b6e3060b327c2711075c"
+  },
+  {
+   "file": "traces/coder-022.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "b3652096c46b556ae7753c3989e642dd7a0ac276ce1652bed20882e831038426"
+  },
+  {
+   "file": "traces/coder-023.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "7b9b612b9359f5b5b440b85d91be7a5895cc07fa3dca6655047288994f78ba36"
+  },
+  {
+   "file": "traces/coder-024.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "e58dbabf22f4310de6033da267bb46159784158b8b296ff11da2d317b1fbbe46"
+  },
+  {
+   "file": "traces/coder-025.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "dd23ca76a539f7acc6d778a47d66cccdbb9463a52edd41d9fbbe9b6a161eb57d"
+  },
+  {
+   "file": "traces/coder-026.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "274495422eae233bdd9e181452d38d4df570f66984d4cccf39a83bfb92809eaa"
+  },
+  {
+   "file": "traces/coder-027.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "e3f7b5a5b25c3a0e48f34268d2498cc881f797865882146952e4989913f260c1"
+  },
+  {
+   "file": "traces/coder-028.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "55c92711a99372dd1f4f5e49013159b0abf93491c6509e8ab37c70c95152a589"
+  },
+  {
+   "file": "traces/coder-029.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "7d89ece5f4d420e05f6f7c3dd4404bec2dd4165f946499dd1a55ef119c85f044"
+  },
+  {
+   "file": "traces/coder-030.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "582bfeac7ba7ad3e5ce251130928270119658f1751b595cda3ed68984e932861"
+  },
+  {
+   "file": "traces/coder-031.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "522ba1013feaaf66987e1759e47db9a804433aee50316376d64bd894eebfb023"
+  },
+  {
+   "file": "traces/coder-032.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "e7009140ecbb97d2a95d3b76efb54e4272f31247147153c84af1c58da63b3f31"
+  },
+  {
+   "file": "traces/coder-033.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "0f3de0b999ef08588b5fcd001896988293b6f85f2c5c6e12cf0ed4afcf836634"
+  },
+  {
+   "file": "traces/coder-034.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "343afd89461ad41c6c7ed98fa08b2abf01d2f77e793dc0835ea4fb9aff4d3fe1"
+  },
+  {
+   "file": "traces/coder-035.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "f6a428da2cf22cc834d00bdabd684e50601c5c17bbdec855de48c21835b48ff2"
+  },
+  {
+   "file": "traces/coder-036.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "9f00f8cf567a800d2d26433592e183f2de15df9d448215254989f3165f1c4846"
+  },
+  {
+   "file": "traces/coder-037.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "0ce6b72a958405d0cbb8b63000e529a5ff1c6e7e216cbdf1a92c60e41dfe0d05"
+  },
+  {
+   "file": "traces/coder-038.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "b00c8c8a20fbad37900daa9e27ea1514a39b691499996a3a7cde25d46f157307"
+  },
+  {
+   "file": "traces/coder-039.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "edac73d8f135b75d8f0381c926e7a17f308ebe9fee95e157af6638c36c802061"
+  },
+  {
+   "file": "traces/coder-040.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "7eee6a53bb098705a4b16e360e4fed63f58911c5dc6d38fbd3fa4ce2bd4d4be4"
+  },
+  {
+   "file": "traces/coder-041.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "9bd12346068888a2a5ced0bf7c6e5ef843405a2372fae1a69d6e67715ad92562"
+  },
+  {
+   "file": "traces/coder-042.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "b4d86c065b8ef9e623e209aca616ec56d375a553def7625766c2ffd5cafb6e3c"
+  },
+  {
+   "file": "traces/coder-043.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "b290c91be60abacd284a80e38a8ad708d5c7d6b1a4f1a87f0eda3ba48c961be8"
+  },
+  {
+   "file": "traces/coder-044.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "bb75b137acfa71099a4271fb2530c14f849b30daa838896c7808da0a49bab1dc"
+  },
+  {
+   "file": "traces/coder-045.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "ef9b7141cb8d0a0cfa4697120f2572401a525cdc25d6ee510638b03299a0a037"
+  },
+  {
+   "file": "traces/coder-046.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "6662c8fe9e3a811e064a240f4544aa4794377e0c4e48a4573ac87260e94f9d85"
+  },
+  {
+   "file": "traces/coder-047.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "1c982a867b8af4194813f0dafb861a327a36612440554cabe1030ddab5cea758"
+  },
+  {
+   "file": "traces/coder-048.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "54fb40bbb11cd2dc50eed3018bae24a65ba99bc18c9e8aa6e5c9f7b6b0d34836"
+  },
+  {
+   "file": "traces/coder-049.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "2a1a3a431d96f3b158d2748a6f5c79795505f75bf366892f115fec98b46fe8cf"
+  },
+  {
+   "file": "traces/coder-050.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "6c4983a8fcfbb8dc8808153e7677e65411311df4aa74312fe8c1da2d1d65ddd0"
+  },
+  {
+   "file": "traces/coder-051.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "e0a3da57eb12faf2a3e0700e89e2fa2dc62d1d6fff2b0df9af7a3a38c32e510d"
+  },
+  {
+   "file": "traces/coder-052.parquet",
+   "rows": 256,
+   "unique_rows_added": 256,
+   "sha256": "0dd615bec8b183ede639b7fdc0a0b9dfa9f9aa16249cfa23c048b82287d66fad"
+  },
+  {
+   "file": "traces/coder-053.parquet",
+   "rows": 32,
+   "unique_rows_added": 32,
+   "sha256": "2902c49fddd8ef684a7c16946eb374a1c7814ce0f302ed4fb1dd5bcde226c692"
+  },
+  {
+   "file": "traces/instruct-001.parquet",
+   "rows": 788,
+   "unique_rows_added": 788,
+   "sha256": "8e9fe6d3bfa56d06cfcab83891eef3a10a7d1c9360dec8ae97e58d5e042df40a"
+  },
+  {
+   "file": "traces/instruct-002.parquet",
+   "rows": 788,
+   "unique_rows_added": 788,
+   "sha256": "fb4c36416302e91c5f813dc81c401aa232a9ebe1ad186dade242d3cdd39b6408"
+  },
+  {
+   "file": "traces/instruct-003.parquet",
+   "rows": 788,
+   "unique_rows_added": 788,
+   "sha256": "f2d9e182011f7b59b87b50858581fa216ddef71c627fe90089fa2b680384fc2f"
+  },
+  {
+   "file": "traces/instruct-004.parquet",
+   "rows": 788,
+   "unique_rows_added": 788,
+   "sha256": "1dbffbb4f21ac018abed9e0796d7393803cf07c3d20b8362ed2b652fa4f07d62"
+  },
+  {
+   "file": "traces/instruct-005.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "c6181f8aeb7375fb964cd71365a6cd9d3943486ef65786e8ba09a849ba543443"
+  },
+  {
+   "file": "traces/instruct-006.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "5a17b0e979e7333463c1a0153c5c4fd04e44b4ae7813346869726ef7d618f9a8"
+  },
+  {
+   "file": "traces/instruct-007.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "0610da581bf5fc923b955794d801363ed42be013412f89874c57fe5bcc20c1fb"
+  },
+  {
+   "file": "traces/instruct-008.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "5e64b710d270d5251b849ba44db0e5ee64db2f72b66aa56ab22b903a531c533e"
+  },
+  {
+   "file": "traces/instruct-009.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "b71e01e6e7013375d0e3ab8f4026a6c9063709e7adcb6622be205ccdf476cc71"
+  },
+  {
+   "file": "traces/instruct-010.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "1407b080c1bf1328b6ad9fb9f741ab0243a1b2315bfed175062b46e65e00f66c"
+  },
+  {
+   "file": "traces/instruct-011.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "0789df0dccde80f41d55aa587c6fa78733c60e1f517de70d8a41c01939b616b1"
+  },
+  {
+   "file": "traces/instruct-012.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "ac86898d5bf0950dc456505cb5093455faacbbb54fa51dfcc305685587f2fca7"
+  },
+  {
+   "file": "traces/instruct-013.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "d9e7978ff051a9dc805a966aed228a3e32e76985494581abb8ef7176b76b963a"
+  },
+  {
+   "file": "traces/instruct-014.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "383a84a011ce08e793ddf73fc5c7b4c719d84579e3a8dff11b8770d6ccc88c13"
+  },
+  {
+   "file": "traces/instruct-015.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "8f7063171ada285b75d3e04dcb11ab296f029bb60bb9a5b8bf5c2b48bd878d47"
+  },
+  {
+   "file": "traces/instruct-016.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "ac9286203ce634b4175a749907e73bfca57cb93b3ff4d493b31e9aa94626f0ba"
+  },
+  {
+   "file": "traces/instruct-017.parquet",
+   "rows": 784,
+   "unique_rows_added": 784,
+   "sha256": "092b603924d39fac200a4f3228c0f3085850d2ea399b67e2eda1dc1102f1e7c0"
+  }
+ ],
+ "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-a3b-instruct-2507": {
+   "trace_distributions": "plots/qwen3-30b-a3b-instruct-2507-trace-distributions.png",
+   "trace_outcomes": "plots/qwen3-30b-a3b-instruct-2507-trace-outcomes.png",
+   "trace_commands": "plots/qwen3-30b-a3b-instruct-2507-trace-commands.png"
+  },
+  "qwen3-coder-30b-a3b-instruct": {
+   "trace_distributions": "plots/qwen3-coder-30b-a3b-instruct-trace-distributions.png",
+   "trace_outcomes": "plots/qwen3-coder-30b-a3b-instruct-trace-outcomes.png",
+   "trace_commands": "plots/qwen3-coder-30b-a3b-instruct-trace-commands.png"
+  }
+ },
+ "metric_implementation_sha256": "ffa2ef63f6c84b193cd50503ece64f6819abc82b894c041ae9c7fb601ef33fe1",
+ "outcome_implementation_sha256": "570e8bdb6930588cb87b9f3728518cae549fd20da595808a2ecef7fed75dddc0",
+ "plotting": {
+  "implementation_sha256": "3e3bdc00f8b2100de4f96bdb48ce6fd87eff518a6bb380f9b07c3da352ad7db6"
+ },
+ "outcome_policy": "exclusive-outcomes-v1"
+}