FineEnvs v3: 515 newly solved tasks and all 1,998 Qwen/Qwen3.8-27B attempts

Adds `fineenvs_v3/` only; nothing else in the repository changes. Please review the README text before merging.

- Evaluated the 999 FineEnvs v2 tasks without a successful attempt: Qwen/Qwen3.8-27B (thinking on), 2 attempts per task, Terminus-2, 32,768-token context, 8,192 output tokens per call.
- 515 newly kept tasks (`tasks.parquet`, disjoint from the 4,001 v2 tasks; 4,516 combined); 484 still without success (`archive.parquet`). Pass@1 44.89%, pass@2 51.55%; success counts 0/2: 484, 1/2: 133, 2/2: 382.
- `traces/`: all 1,998 attempts (successful and failed) with conversations and result records; 6 attempts without a verifier score (task memory limit), counted as unsuccessful.
- `trace_stats/`, `plots/`, `metadata/` (selection, results, settings audit of all 1,998 saved configurations, contract, stage reports, release verification) and `patch.py`, which reproduces both task files from the pinned v2 archive.
- Task payloads are unchanged from the evaluated v2 archive rows (content hashes verified).

fineenvs_v3/README.md

@@ -0,0 +1,99 @@
+# FineEnvs v3
+
+## Dataset stats
+
+- **Tasks:** 999 had no successful attempt in v2; all 999 evaluated for v3; 515 newly kept; 484 still without success.
+- **Teacher evaluation:** Qwen/Qwen3.8-27B (thinking on), 2 attempts per task (pass@2).
+- **Coverage:** 999 tasks evaluated; 1,998 evaluation attempts.
+- **Saved conversations:** 1,998 of 1,998 published attempt records, including successful and unsuccessful attempts.
+
+We keep tasks with at least one verifier-successful attempt from the new model. `tasks.parquet` contains only the 515 newly kept tasks; the 4,001 tasks already kept in v2 remain separate. Task payloads are unchanged from the evaluated v2 archive rows.
+
+## Evaluation settings
+
+| Model | Agent | Thinking | Attempts/task | Temperature | Top-p | Top-k | Context tokens | Intended output tokens/call | Other sampling |
+| --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
+| Qwen/Qwen3.8-27B | Terminus-2 | On | 2 | 1.0 | 0.95 | 20 | 32,768 | 8,192 | min_p=0; presence_penalty=0; repetition_penalty=1 |
+
+Context includes prompt and response for one model call. The output budget includes thinking tokens; neither is a whole-attempt token total.
+
+**Time limits:** agent 600 seconds for 1,972 attempts and 900 seconds for 26, set by task; verifier 120 seconds for all attempts. No separate total trial timeout is configured. The task memory limit is 972.8 MB for 1,972 attempts and 3,891.2 MB for 26.
+
+**Output-limit evidence:** All 1,998 saved configurations contain an explicit 8,192-token request cap. HTTP payloads and server enforcement were not audited.
+
+**Network:** All tasks declare internet access. The runs requested an isolated container network, but Helma could not create network namespaces, so every trial used the host network. Agents had internet access, as the tasks allow, but containers were not network-isolated from the host.
+
+**Stage-1 checks:** All 999 tasks passed the default training-profile checks before evaluation; the AI-detection check is excluded by policy. Repository-history and timestamp leak checks found no git repository in any task environment. In 49 evaluated tasks (22 kept), the instruction names an input file by a relative path that was not found under `/home/user/input`, `/workdir` or `/workspace` before the agent started. These are advisory findings; the tasks were evaluated unchanged.
+
+## Results
+
+The tables and plot cover all evaluated tasks, before selecting tasks for publication. Only the new v3 evaluation is included.
+
+| Successful attempts | Qwen/Qwen3.8-27B (999 tasks) |
+| --- | ---: |
+| 0/2 | 484 |
+| 1/2 | 133 |
+| 2/2 | 382 |
+
+| Model | Evaluated tasks | Pass@1 | Pass@2 |
+| --- | ---: | ---: | ---: |
+| Qwen/Qwen3.8-27B | 999 | 44.89% | 51.55% |
+
+Pass@k uses the sampling-without-replacement estimator from the recorded per-task success counts. At the full attempt budget, it is the fraction of evaluated tasks with at least one observed success. 897 of 1,998 attempts succeeded.
+
+Six attempts on five tasks have no verifier score. Each reached the 972.8 MB task memory limit, five while installing TensorFlow and one while running XGBoost and pandas code, and the agent's terminal stopped responding (`TmuxBatchProtocolError`). The recorded policy counts these as unsuccessful (reward 0); their raw verifier rewards remain null. Two of the five tasks were solved by their other attempt.
+
+![Pass rates across all evaluated tasks](plots/readme-pass-at-k.png)
+
+## Files and usage
+
+`tasks.parquet` contains 515 newly kept tasks only. Tasks already kept in v2 are not duplicated.
+
+`traces/` contains all 1,998 attempts of the v3 evaluation, including unsuccessful attempts and tasks not kept. Rows retain task/model/attempt identifiers, rewards, `retained_v3` task membership, the full conversation and the raw result record.
+
+Success-count columns: `qwen38_successes` and `qwen38_attempts`. All other columns follow the v1 schema used in v2.
+
+`archive.parquet` preserves the 484 tasks without a success in either evaluation, with the archive reason. `patch.py` reproduces both files from the pinned v2 archive.
+
+```python
+from huggingface_hub import hf_hub_download
+import pyarrow.parquet as pq
+
+path = hf_hub_download(
+    "FWeindel/validated-tasks",
+    "fineenvs_v3/tasks.parquet",
+    repo_type="dataset",
+)
+# Select kept tasks with exactly one observed success.
+tasks = pq.read_table(path, filters=[("qwen38_successes", "=", 1)])
+```
+
+Combine the 4,001 tasks in `fineenvs_v2/tasks.parquet` with these 515 newly kept tasks for 4,516 unique tasks. Allow for the added columns and retain each source file's image manifest when staging images.
+
+```python
+import pyarrow as pa
+
+def load(version):
+    path = hf_hub_download(
+        "FWeindel/validated-tasks",
+        f"fineenvs_{version}/tasks.parquet",
+        repo_type="dataset",
+    )
+    return pq.read_table(path)
+
+v2, v3 = load("v2"), load("v3")
+assert set(v2["path"].to_pylist()).isdisjoint(v3["path"].to_pylist())
+combined = pa.concat_tables([v2, v3], promote_options="permissive")
+```
+
+V3 tasks are harder relative to the v2 evaluation, not necessarily for every model. Teacher success does not establish completion of all current datasource-validation checks.
+
+## Trace statistics
+
+Of 1,998 recorded attempts, 1,998 have saved conversations and 0 do not. 6 attempts have no raw verifier reward. Missing measurements remain missing.
+
+Agent timeouts and verifier outcomes are separate: an attempt can time out and still receive a successful verifier grade.
+
+The export includes successful and unsuccessful attempts on both kept and unselected tasks. Conversation availability is not a training-quality acceptance claim; no additional conversation quality filter is applied to this export. Historical attempts remain in v2.
+
+`trace_stats/README.md` contains detailed turns, token usage, context, runtime, outcome and command distributions. Token totals sum calls across an attempt and are not the context window.

fineenvs_v3/archive.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7a1473e1851abc0b3648f266ee009ada1725219b97bedb919cc894a3fe46734
+size 6820682614

fineenvs_v3/metadata/contract.json

@@ -0,0 +1,1952 @@
+{
+  "schema_version": 3,
+  "stage_layout": "environment-first-v2",
+  "created_at": "2026-10-26T17:35:56.983354+00:00",
+  "dataset": {
+    "source": "FWeindel/validated-tasks/fineenvs_v2",
+    "reference_solutions": null,
+    "instruction_suffixes": null,
+    "revision": "6dcbab71a3926fe61a5d630a8e7f64edb16f1a54",
+    "automation_changes": {},
+    "source_files": [
+      {
+        "path": "/hnvme/workspace/y500bb12-optiagent/fineenvs-v2-20261026/publication/fineenvs_v2/archive.parquet",
+        "sha256": "fbfee6fd39341d24057ff4b355de7a0c4b91923faa17fb4ea8bdded42b6c58b8"
+      }
+    ],
+    "preselection": null,
+    "normalization": null,
+    "selection": {
+      "method": "all",
+      "limit": null,
+      "task_id_range_inclusive": null,
+      "order": "lexicographic IDs for ranges; otherwise sorted Parquet paths then row order, or sorted task directories",
+      "implementation": "validation/contract.py:inventory (hash recorded in implementation)"
+    }
+  },
+  "stages": [
+    1,
+    6,
+    7
+  ],
+  "success_criteria": {
+    "minimum_tasks": 999,
+    "oracle_reward": 1,
+    "nop_reward": 0,
+    "reward_key": "reward",
+    "attempts": 2,
+    "static_checks": [
+      "check-compose-host-binds.sh",
+      "check-dockerfile-platform.sh",
+      "check-dockerfile-sanity.sh",
+      "check-nproc.sh",
+      "check-pip-pinning.sh",
+      "check-resource-sizes.sh",
+      "check-task-timeout.sh",
+      "check-task-toml-schema.py",
+      "check-trial-network-fetch.sh",
+      "check-verifier-tooling-baked.sh"
+    ],
+    "static_checks_by_stage": {
+      "1": [
+        "check-compose-host-binds.sh",
+        "check-dockerfile-platform.sh",
+        "check-dockerfile-sanity.sh",
+        "check-nproc.sh",
+        "check-pip-pinning.sh",
+        "check-resource-sizes.sh",
+        "check-task-timeout.sh",
+        "check-task-toml-schema.py",
+        "check-trial-network-fetch.sh",
+        "check-verifier-tooling-baked.sh"
+      ]
+    },
+    "static_exclusions": {
+      "check-task-changelog.sh": "training tasks do not require Terminal-Bench PR changelogs",
+      "check-test-sh-sanity.sh": "training permits shared system Python and image-installed verifier dependencies; uv is not required, and stages 3/3 check execution",
+      "check-separate-verifier.sh": "training permits grading in the agent environment, including installed packages and changed system state; separate verification is optional",
+      "check-allow-internet.sh": "network policy is declared by the task and checked at runtime",
+      "check-no-allow-internet-true.sh": "explicit task network policies are allowed",
+      "check-pytest-version.sh": "training tasks may use other pinned pytest/CTRF versions",
+      "check-gpu-types.sh": "Modal GPU naming conventions do not apply to the Helma execution profile",
+      "check-canary.sh": "training tasks do not require a benchmark canary string",
+      "check-instruction-headings.sh": "Markdown headings are allowed in training instructions",
+      "check-task-fields.sh": "training tasks do not require TB author GitHub metadata, taxonomy or explanation sections",
+      "check-task-package-name.sh": "training task package names need not use the terminal-bench namespace",
+      "check-task-slug.sh": "training task names may contain more than three hyphen-separated tokens",
+      "rubric_review.py": "LLM rubric review is opt-in via stage 5, never a default static check"
+    },
+    "conditional_checks": {
+      "check_ai_detection.py": "only with GPTZERO_API_KEY; missing key or upstream API-unavailable skip is non-failing"
+    },
+    "failure_policy": "collect all findings; fail acceptance on any selected failure/error/skip or insufficient coverage"
+  },
+  "execution_profile": {
+    "operation": "validation",
+    "dependency_archives": null,
+    "instruction_path_search_roots": [],
+    "agent_settings": {
+      "settings": {
+        "model_name": null,
+        "max_turns": null,
+        "parser_name": "json",
+        "api_base": null,
+        "temperature": 1.0,
+        "reasoning_effort": null,
+        "collect_rollout_details": false,
+        "session_id": null,
+        "enable_summarize": true,
+        "proactive_summarization_threshold": 8000,
+        "max_thinking_tokens": null,
+        "model_info": null,
+        "trajectory_config": null,
+        "trajectory_dump_cadence": null,
+        "tmux_pane_width": 160,
+        "tmux_pane_height": 40,
+        "store_all_messages": false,
+        "enable_episode_logging": false,
+        "interleaved_thinking": false,
+        "suppress_max_turns_warning": false,
+        "use_responses_api": false,
+        "llm_backend": "litellm",
+        "llm_kwargs": null,
+        "llm_call_kwargs": {
+          "max_tokens": 8192
+        },
+        "extra_body": {
+          "top_p": 0.95,
+          "top_k": 20,
+          "min_p": 0,
+          "presence_penalty": 0,
+          "repetition_penalty": 1.0,
+          "chat_template_kwargs": {
+            "enable_thinking": true
+          }
+        },
+        "extra_env": null,
+        "turn_callback": null,
+        "batch_turn_execs": true,
+        "batch_exec_timeout_margin_sec": 180.0,
+        "max_tokens": 8192
+      },
+      "defaults_resolved": true,
+      "source_sha256": {
+        "terminus_2.py": "256df8b60461c9326468f843e5c3a0f483d03b473a47d6b5a1d442ab0b6d48cd",
+        "terminus_json_plain_parser.py": "b20ad9b8931d4f92c31aa588126a342d9cc6a59396f3ef21243bd077895d9022",
+        "terminus_xml_plain_parser.py": "11ff85d848d4ada5874644b5113b08dbcc53de81deb6ba4e014ca6fce92435fa",
+        "templates/terminus-json-plain.txt": "89a3dc3a15752b748a99fb907c9251ab31a464e13386583fb48476b078b54a34",
+        "templates/terminus-xml-plain.txt": "721c0ada54921cbb7bb9f817f2306c79ff4ba3381d08be7eb97fbeb0d99d298d",
+        "templates/timeout.txt": "32bf9aa7b157a6a0e0ba0d33b81e59dd67204276f158afc1fe6cb684ced68872"
+      }
+    },
+    "backend": "apptainer",
+    "architecture": "x86_64",
+    "partition_policy": "h200",
+    "allocation_policy": "auto: CPU for external models; h200 for local models or explicit GPUs; h200 fallback if CPU partition has no usable nodes",
+    "scheduler_target": "auto",
+    "submission_host": "helma1",
+    "network": "isolated",
+    "network_guarantee": "isolation requested; existing bridge may fall back to host network; offline isolation is NOT established",
+    "privileges": "fakeroot requested, host UID unprivileged; bridge may fall back without fakeroot",
+    "mount_policy": "per-trial workdir, tests, logs, home and tmp; site hostfs/bind-paths/cwd disabled; baked verifier tests preserved; explicit read-only host DNS bind only in host network mode",
+    "profile_scope": "declared bridge policy including stated fallbacks; does not certify offline or rootless behavior",
+    "dependency_policy": "execute shipped Dockerfiles; mutable tags and unpinned installs are not rewritten; reuse content-keyed image cache unless force_build; installed agents may download dependencies",
+    "environment_kwargs": {},
+    "build_retry_policy": "after initial pass, retry failed tasks three times in fresh environments; require all three to pass; skip subsequent oracle/NOP on failure",
+    "preparation_measurement": {
+      "review_setup": null,
+      "runs": 1,
+      "mean_target_seconds": null,
+      "median_target_seconds": null,
+      "verifier_review": null,
+      "max_seconds": null,
+      "startup_policy": "review records startup separately, bounded by each environment.build_timeout_sec; startup failures fail validation",
+      "scope": "review preparation excludes container startup; fresh task preparation; review also measures verifier transfers and tests/setup.sh separately; excludes image builds, inspection and grading"
+    },
+    "force_build": false,
+    "trial_cpus": null,
+    "trial_memory_mb": null,
+    "agent": "terminus-2",
+    "model": null,
+    "review_agent": null,
+    "review_rubric": "terminal-bench",
+    "review_format": null,
+    "review_runs": 2,
+    "review_retries": 2,
+    "review_resume": true,
+    "review_seed": 42,
+    "review_evidence": "image",
+    "review_criteria": null,
+    "review_model": null,
+    "analysis_agent": null,
+    "analysis_model": null,
+    "review_defaults_source": null,
+    "agent_kwargs": {
+      "temperature": 1.0,
+      "max_tokens": 8192,
+      "llm_call_kwargs": {
+        "max_tokens": 8192
+      },
+      "extra_body": {
+        "top_p": 0.95,
+        "top_k": 20,
+        "min_p": 0,
+        "presence_penalty": 0,
+        "repetition_penalty": 1.0,
+        "chat_template_kwargs": {
+          "enable_thinking": true
+        }
+      }
+    },
+    "review_local": false,
+    "serve_weights": "/hnvme/workspace/y500bb12-optiagent/models/Qwen3.8-27B",
+    "serve_context": 32768,
+    "serve_replicas": 1,
+    "local_model_assets": {
+      "spec": {
+        "name": "Qwen3.8-27B",
+        "gpus": 1,
+        "sampling": {
+          "temperature": 0.7,
+          "top_p": 0.8,
+          "top_k": 20,
+          "presence_penalty": 1.5,
+          "repetition_penalty": 1.0
+        },
+        "context": 32768,
+        "max_output_tokens": 32767,
+        "hf_repo": "Qwen/Qwen3.8-27B",
+        "thinking": true,
+        "extra_args": [
+          "--language-model-only"
+        ],
+        "max_tensor_parallel": 0,
+        "dtype": "bfloat16",
+        "reasoning_parser": "qwen3",
+        "tool_call_parser": "qwen3_xml",
+        "weights_gb": 56
+      },
+      "weights": "/hnvme/workspace/y500bb12-optiagent/models/Qwen3.8-27B",
+      "revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+      "assets_sha256": {
+        "config.json": "191e0af232104ed8b65258cf3fb2b842e288008baca7633c11b82a1ac7203aab",
+        "tokenizer_config.json": "b11349aafa7cdc6a320767cf7ceb29ed82f7eda5d65e8e0819e76f0ce947bf27",
+        "chat_template.jinja": "c3cf9e34abf4f9e36c2d72165aa9c132d3e2a725b6c2586aaa3a8af9d7a81041",
+        "generation_config.json": "e70c136c1b78ddc1fb0905bac8e733a4dc448d4f852a5dd75143fffc70be550e"
+      },
+      "runtime": {
+        "python": "3.12.14",
+        "requirements_sha256": "fb46a4637021c792e0a01bec69d66660ed13cd71ceab444503dcd86b6a098ebc",
+        "packages": {
+          "Jinja2": "3.1.6",
+          "MarkupSafe": "3.0.4",
+          "PyJWT": "2.15.1",
+          "PyYAML": "6.0.3",
+          "Pygments": "2.21.0",
+          "StrEnum": "0.4.15",
+          "agent-detector": "2.0.0",
+          "aiohappyeyeballs": "2.7.1",
+          "aiohttp": "3.14.4",
+          "aiohttp-cors": "0.8.1",
+          "aiosignal": "1.4.0",
+          "annotated-doc": "0.0.5",
+          "annotated-types": "0.8.0",
+          "anthropic": "1.11.0",
+          "anyio": "4.15.1",
+          "apache-tvm-ffi": "0.1.11",
+          "astor": "0.8.1",
+          "attrs": "26.1.0",
+          "blake3": "1.0.10",
+          "boto3": "1.43.108",
+          "botocore": "1.43.108",
+          "cachetools": "7.2.1",
+          "cbor2": "6.1.5",
+          "certifi": "2026.7.22",
+          "cffi": "2.1.1",
+          "charset-normalizer": "3.5.2",
+          "click": "8.5.0",
+          "cloudpickle": "3.1.2",
+          "colorful": "0.5.8",
+          "compressed-tensors": "0.17.0",
+          "contourpy": "1.4.0",
+          "cryptography": "50.0.2",
+          "cuda-bindings": "13.4.3",
+          "cuda-core": "1.2.1",
+          "cuda-pathfinder": "1.8.3",
+          "cuda-python": "13.4.1",
+          "cuda-tile": "1.6.0",
+          "cuda-toolkit": "13.0.3.0",
+          "cycler": "0.12.1",
+          "deprecation": "2.1.0",
+          "depyf": "0.20.0",
+          "detect-installer": "0.2.1",
+          "dill": "0.4.1",
+          "dirhash": "0.5.0",
+          "distlib": "0.4.3",
+          "distro": "1.9.0",
+          "dnspython": "2.8.0",
+          "docstring_parser": "0.18.0",
+          "einops": "0.8.2",
+          "email-validator": "2.3.0",
+          "fastapi": "0.136.3",
+          "fastapi-cli": "0.0.32",
+          "fastapi-cloud-cli": "0.26.0",
+          "fastar": "0.12.0",
+          "fastsafetensors": "0.4.0",
+          "fastuuid": "0.14.0",
+          "filelock": "3.32.7",
+          "flashinfer-python": "0.6.18.post1",
+          "fonttools": "4.66.1",
+          "frozenlist": "1.8.0",
+          "fsspec": "2026.9.0",
+          "google-api-core": "2.41.0",
+          "google-auth": "2.60.0",
+          "googleapis-common-protos": "1.75.5",
+          "grpcio": "1.84.0",
+          "h11": "0.16.0",
+          "h2": "4.4.1",
+          "harbor": "0.8.1",
+          "hf-xet": "1.6.0",
+          "hpack": "4.2.0",
+          "httpcore": "1.0.9",
+          "httpcore2": "2.13.1",
+          "httptools": "0.8.0",
+          "httpx": "0.28.1",
+          "httpx2": "2.13.1",
+          "huggingface_hub": "1.33.0",
+          "humming-kernels": "0.1.12",
+          "hyperframe": "6.1.0",
+          "idna": "3.20",
+          "ijson": "3.5.1",
+          "importlib_metadata": "8.9.0",
+          "iniconfig": "2.3.0",
+          "instanttensor": "0.2.1",
+          "interegular": "0.3.3",
+          "jiter": "0.17.0",
+          "jmespath": "1.1.0",
+          "jsonschema": "4.26.0",
+          "jsonschema-specifications": "2025.9.1",
+          "kiwisolver": "1.5.1",
+          "lark": "1.2.2",
+          "litellm": "1.104.0",
+          "llguidance": "1.7.6",
+          "llvmlite": "0.47.0",
+          "lm-format-enforcer": "0.11.3",
+          "loguru": "0.7.3",
+          "markdown-it-py": "4.2.0",
+          "matplotlib": "3.11.2",
+          "mcp": "2.3.0",
+          "mcp-types": "2.3.0",
+          "mdurl": "0.1.2",
+          "mistral_common": "1.12.0",
+          "ml_dtypes": "0.6.0",
+          "model-hosting-container-standards": "0.1.16",
+          "mpmath": "1.3.0",
+          "msgpack": "1.2.3",
+          "msgspec": "0.22.0",
+          "multidict": "6.9.1",
+          "nccl4py": "0.6.0",
+          "networkx": "3.7",
+          "ninja": "1.13.2",
+          "numba": "0.65.0",
+          "numpy": "2.3.5",
+          "nvidia-cublas": "13.1.1.3",
+          "nvidia-cuda-cccl": "13.3.4.3.1",
+          "nvidia-cuda-crt": "13.4.92",
+          "nvidia-cuda-cupti": "13.0.85",
+          "nvidia-cuda-nvcc": "13.4.92",
+          "nvidia-cuda-nvdisasm": "13.4.92",
+          "nvidia-cuda-nvrtc": "13.0.88",
+          "nvidia-cuda-runtime": "13.0.96",
+          "nvidia-cudnn-cu13": "9.20.0.48",
+          "nvidia-cudnn-frontend": "1.30.0",
+          "nvidia-cufft": "12.0.0.61",
+          "nvidia-cufile": "1.15.1.6",
+          "nvidia-curand": "10.4.0.35",
+          "nvidia-cusolver": "12.0.4.66",
+          "nvidia-cusparse": "12.6.3.3",
+          "nvidia-cusparselt-cu13": "0.8.1",
+          "nvidia-cutlass-dsl": "4.7.1",
+          "nvidia-cutlass-dsl-libs-base": "4.7.1",
+          "nvidia-cutlass-dsl-libs-core": "4.7.1",
+          "nvidia-cutlass-dsl-libs-cu12": "4.7.1",
+          "nvidia-cutlass-dsl-libs-cu13": "4.7.1",
+          "nvidia-ml-py": "13.615.71",
+          "nvidia-nccl-cu13": "2.29.7",
+          "nvidia-nvjitlink": "13.4.92",
+          "nvidia-nvshmem-cu13": "3.4.5",
+          "nvidia-nvtx": "13.0.85",
+          "nvidia-nvvm": "13.4.92",
+          "nvtx": "0.2.15",
+          "openai": "2.54.0",
+          "openai-harmony": "0.0.8",
+          "opencensus": "0.11.4",
+          "opencensus-context": "0.1.3",
+          "opencv-python-headless": "5.0.0.93",
+          "opentelemetry-api": "1.45.1",
+          "opentelemetry-exporter-http-transport": "0.66b1",
+          "opentelemetry-exporter-otlp": "1.45.1",
+          "opentelemetry-exporter-otlp-common": "0.66b1",
+          "opentelemetry-exporter-otlp-proto-common": "1.45.1",
+          "opentelemetry-exporter-otlp-proto-grpc": "1.45.1",
+          "opentelemetry-exporter-otlp-proto-http": "1.45.1",
+          "opentelemetry-exporter-prometheus": "0.66b1",
+          "opentelemetry-proto": "1.45.1",
+          "opentelemetry-sdk": "1.45.1",
+          "opentelemetry-semantic-conventions": "0.66b1",
+          "opentelemetry-semantic-conventions-ai": "0.5.1",
+          "outlines_core": "0.2.14",
+          "packaging": "26.3",
+          "partial-json-parser": "0.2.1.1.post7",
+          "pathlib_abc": "0.5.2",
+          "pathspec": "1.1.1",
+          "pillow": "12.3.0",
+          "pip": "25.0.1",
+          "platformdirs": "4.12.3",
+          "pluggy": "1.6.0",
+          "postgrest": "2.32.0",
+          "prometheus-fastapi-instrumentator": "8.1.0",
+          "prometheus_client": "0.26.0",
+          "propcache": "0.5.4",
+          "proto-plus": "1.29.0",
+          "protobuf": "7.36.2",
+          "psutil": "7.2.2",
+          "py-cpuinfo": "9.0.0",
+          "py-spy": "0.4.2",
+          "pyarrow": "25.0.1",
+          "pyasn1": "0.6.4",
+          "pyasn1_modules": "0.4.2",
+          "pybase64": "1.5.1",
+          "pycountry": "26.2.16",
+          "pycparser": "3.0",
+          "pydantic": "2.13.5",
+          "pydantic-extra-types": "2.11.1",
+          "pydantic-settings": "2.15.0",
+          "pydantic_core": "2.46.5",
+          "pynvvideocodec": "2.0.4",
+          "pyparsing": "3.3.3",
+          "pytest": "9.1.1",
+          "python-dateutil": "2.9.0.post0",
+          "python-discovery": "1.6.1",
+          "python-dotenv": "1.2.4",
+          "python-json-logger": "4.2.0",
+          "python-multipart": "0.0.32",
+          "pyzmq": "27.2.0",
+          "quack-kernels": "0.6.5",
+          "ray": "2.55.1",
+          "realtime": "2.32.0",
+          "referencing": "0.37.0",
+          "regex": "2026.9.29",
+          "requests": "2.34.2",
+          "rich": "15.0.0",
+          "rich-toolkit": "0.20.5",
+          "rignore": "0.8.1",
+          "rpds-py": "2026.9.1",
+          "s3transfer": "0.19.2",
+          "safetensors": "0.8.0",
+          "scantree": "0.0.4",
+          "sentencepiece": "0.2.2",
+          "sentry-sdk": "2.71.0",
+          "setproctitle": "1.3.8",
+          "setuptools": "80.10.2",
+          "shellingham": "1.5.4",
+          "shortuuid": "1.0.13",
+          "six": "1.17.0",
+          "smart_open": "8.0.2",
+          "sniffio": "1.3.1",
+          "sse-starlette": "3.5.0",
+          "starlette": "1.7.0",
+          "storage3": "2.32.0",
+          "supabase": "2.32.0",
+          "supabase-auth": "2.32.0",
+          "supabase-functions": "2.32.0",
+          "supervisor": "4.3.0",
+          "sympy": "1.14.0",
+          "tabulate": "0.10.0",
+          "tenacity": "9.1.4",
+          "tiktoken": "0.14.0",
+          "tilelang": "0.1.12",
+          "tokenizers": "0.23.2",
+          "tokenspeed-mla": "0.1.8",
+          "tokenspeed-triton": "3.8.10.post20260920",
+          "toml": "0.10.2",
+          "torch": "2.13.0",
+          "torch_c_dlpack_ext": "0.1.5",
+          "torchaudio": "2.11.0",
+          "torchcodec": "0.17.0",
+          "torchvision": "0.28.0",
+          "tqdm": "4.70.1",
+          "transformers": "5.19.0",
+          "triton": "3.7.1",
+          "truststore": "0.10.4",
+          "typer": "0.27.3",
+          "typing-inspection": "0.4.4",
+          "typing_extensions": "4.16.0",
+          "universal_pathlib": "0.3.10",
+          "urllib3": "2.8.0",
+          "uvicorn": "0.54.0",
+          "uvloop": "0.23.0",
+          "virtualenv": "21.14.5",
+          "vllm": "0.30.0",
+          "watchfiles": "1.3.0",
+          "websockets": "15.0.1",
+          "wrapt": "2.5.0",
+          "xgrammar": "0.2.8",
+          "yarl": "1.25.1",
+          "z3-solver": "4.15.4.0",
+          "zipp": "4.1.1"
+        },
+        "vllm_required": "0.30.0"
+      },
+      "scope": "configuration hashes and recorded model revision; weight shards are not rehashed"
+    },
+    "serve_model": "qwen38-27b",
+    "api_base": null,
+    "validation_container_reuse": "fresh instance per phase",
+    "nop_setup_policy": "one NOP trial per attempt after environment preparation; no pre-setup NOP",
+    "task_setup_budget_policy": "Slurm image preparation precedes task timers; container startup, detected task setup and setup-files upload share environment.build_timeout_sec; direct runs without preparation also include image build",
+    "container_start_concurrency": 4,
+    "image_preparation": {
+      "memory_mb": 8192,
+      "cpus": 4,
+      "concurrency": 2,
+      "timeout_sec": 3600
+    },
+    "container_start_interval": 0,
+    "static_concurrency": 4,
+    "static_concurrency_policy": "cap at allocated CPU budget; no separate verifier reservation",
+    "concurrency": 4,
+    "concurrency_policy": "cap at allocation CPU budget including separate verifiers; record effective value in resources.json",
+    "cpus": 32,
+    "gpus": null,
+    "serve_cpus": 4,
+    "serve_memory_mb": 32768,
+    "max_num_seqs": 8
+  },
+  "runtime_dependencies": {
+    "python": "3.12.14",
+    "packages": [
+      [
+        "Jinja2",
+        "3.1.6"
+      ],
+      [
+        "MarkupSafe",
+        "3.0.4"
+      ],
+      [
+        "PyJWT",
+        "2.15.1"
+      ],
+      [
+        "PyYAML",
+        "6.0.3"
+      ],
+      [
+        "Pygments",
+        "2.21.0"
+      ],
+      [
+        "StrEnum",
+        "0.4.15"
+      ],
+      [
+        "agent-detector",
+        "2.0.0"
+      ],
+      [
+        "aiohappyeyeballs",
+        "2.7.1"
+      ],
+      [
+        "aiohttp",
+        "3.14.4"
+      ],
+      [
+        "aiohttp-cors",
+        "0.8.1"
+      ],
+      [
+        "aiosignal",
+        "1.4.0"
+      ],
+      [
+        "annotated-doc",
+        "0.0.5"
+      ],
+      [
+        "annotated-types",
+        "0.8.0"
+      ],
+      [
+        "anthropic",
+        "1.11.0"
+      ],
+      [
+        "anyio",
+        "4.15.1"
+      ],
+      [
+        "apache-tvm-ffi",
+        "0.1.11"
+      ],
+      [
+        "astor",
+        "0.8.1"
+      ],
+      [
+        "attrs",
+        "26.1.0"
+      ],
+      [
+        "blake3",
+        "1.0.10"
+      ],
+      [
+        "boto3",
+        "1.43.108"
+      ],
+      [
+        "botocore",
+        "1.43.108"
+      ],
+      [
+        "cachetools",
+        "7.2.1"
+      ],
+      [
+        "cbor2",
+        "6.1.5"
+      ],
+      [
+        "certifi",
+        "2026.7.22"
+      ],
+      [
+        "cffi",
+        "2.1.1"
+      ],
+      [
+        "charset-normalizer",
+        "3.5.2"
+      ],
+      [
+        "click",
+        "8.5.0"
+      ],
+      [
+        "cloudpickle",
+        "3.1.2"
+      ],
+      [
+        "colorful",
+        "0.5.8"
+      ],
+      [
+        "compressed-tensors",
+        "0.17.0"
+      ],
+      [
+        "contourpy",
+        "1.4.0"
+      ],
+      [
+        "cryptography",
+        "50.0.2"
+      ],
+      [
+        "cuda-bindings",
+        "13.4.3"
+      ],
+      [
+        "cuda-core",
+        "1.2.1"
+      ],
+      [
+        "cuda-pathfinder",
+        "1.8.3"
+      ],
+      [
+        "cuda-python",
+        "13.4.1"
+      ],
+      [
+        "cuda-tile",
+        "1.6.0"
+      ],
+      [
+        "cuda-toolkit",
+        "13.0.3.0"
+      ],
+      [
+        "cycler",
+        "0.12.1"
+      ],
+      [
+        "deprecation",
+        "2.1.0"
+      ],
+      [
+        "depyf",
+        "0.20.0"
+      ],
+      [
+        "detect-installer",
+        "0.2.1"
+      ],
+      [
+        "dill",
+        "0.4.1"
+      ],
+      [
+        "dirhash",
+        "0.5.0"
+      ],
+      [
+        "distlib",
+        "0.4.3"
+      ],
+      [
+        "distro",
+        "1.9.0"
+      ],
+      [
+        "dnspython",
+        "2.8.0"
+      ],
+      [
+        "docstring_parser",
+        "0.18.0"
+      ],
+      [
+        "einops",
+        "0.8.2"
+      ],
+      [
+        "email-validator",
+        "2.3.0"
+      ],
+      [
+        "fastapi",
+        "0.136.3"
+      ],
+      [
+        "fastapi-cli",
+        "0.0.32"
+      ],
+      [
+        "fastapi-cloud-cli",
+        "0.26.0"
+      ],
+      [
+        "fastar",
+        "0.12.0"
+      ],
+      [
+        "fastsafetensors",
+        "0.4.0"
+      ],
+      [
+        "fastuuid",
+        "0.14.0"
+      ],
+      [
+        "filelock",
+        "3.32.7"
+      ],
+      [
+        "flashinfer-python",
+        "0.6.18.post1"
+      ],
+      [
+        "fonttools",
+        "4.66.1"
+      ],
+      [
+        "frozenlist",
+        "1.8.0"
+      ],
+      [
+        "fsspec",
+        "2026.9.0"
+      ],
+      [
+        "google-api-core",
+        "2.41.0"
+      ],
+      [
+        "google-auth",
+        "2.60.0"
+      ],
+      [
+        "googleapis-common-protos",
+        "1.75.5"
+      ],
+      [
+        "grpcio",
+        "1.84.0"
+      ],
+      [
+        "h11",
+        "0.16.0"
+      ],
+      [
+        "h2",
+        "4.4.1"
+      ],
+      [
+        "harbor",
+        "0.8.1"
+      ],
+      [
+        "hf-xet",
+        "1.6.0"
+      ],
+      [
+        "hpack",
+        "4.2.0"
+      ],
+      [
+        "httpcore",
+        "1.0.9"
+      ],
+      [
+        "httpcore2",
+        "2.13.1"
+      ],
+      [
+        "httptools",
+        "0.8.0"
+      ],
+      [
+        "httpx",
+        "0.28.1"
+      ],
+      [
+        "httpx2",
+        "2.13.1"
+      ],
+      [
+        "huggingface_hub",
+        "1.33.0"
+      ],
+      [
+        "humming-kernels",
+        "0.1.12"
+      ],
+      [
+        "hyperframe",
+        "6.1.0"
+      ],
+      [
+        "idna",
+        "3.20"
+      ],
+      [
+        "ijson",
+        "3.5.1"
+      ],
+      [
+        "importlib_metadata",
+        "8.9.0"
+      ],
+      [
+        "iniconfig",
+        "2.3.0"
+      ],
+      [
+        "instanttensor",
+        "0.2.1"
+      ],
+      [
+        "interegular",
+        "0.3.3"
+      ],
+      [
+        "jiter",
+        "0.17.0"
+      ],
+      [
+        "jmespath",
+        "1.1.0"
+      ],
+      [
+        "jsonschema",
+        "4.26.0"
+      ],
+      [
+        "jsonschema-specifications",
+        "2025.9.1"
+      ],
+      [
+        "kiwisolver",
+        "1.5.1"
+      ],
+      [
+        "lark",
+        "1.2.2"
+      ],
+      [
+        "litellm",
+        "1.104.0"
+      ],
+      [
+        "llguidance",
+        "1.7.6"
+      ],
+      [
+        "llvmlite",
+        "0.47.0"
+      ],
+      [
+        "lm-format-enforcer",
+        "0.11.3"
+      ],
+      [
+        "loguru",
+        "0.7.3"
+      ],
+      [
+        "markdown-it-py",
+        "4.2.0"
+      ],
+      [
+        "matplotlib",
+        "3.11.2"
+      ],
+      [
+        "mcp",
+        "2.3.0"
+      ],
+      [
+        "mcp-types",
+        "2.3.0"
+      ],
+      [
+        "mdurl",
+        "0.1.2"
+      ],
+      [
+        "mistral_common",
+        "1.12.0"
+      ],
+      [
+        "ml_dtypes",
+        "0.6.0"
+      ],
+      [
+        "model-hosting-container-standards",
+        "0.1.16"
+      ],
+      [
+        "mpmath",
+        "1.3.0"
+      ],
+      [
+        "msgpack",
+        "1.2.3"
+      ],
+      [
+        "msgspec",
+        "0.22.0"
+      ],
+      [
+        "multidict",
+        "6.9.1"
+      ],
+      [
+        "nccl4py",
+        "0.6.0"
+      ],
+      [
+        "networkx",
+        "3.7"
+      ],
+      [
+        "ninja",
+        "1.13.2"
+      ],
+      [
+        "numba",
+        "0.65.0"
+      ],
+      [
+        "numpy",
+        "2.3.5"
+      ],
+      [
+        "nvidia-cublas",
+        "13.1.1.3"
+      ],
+      [
+        "nvidia-cuda-cccl",
+        "13.3.4.3.1"
+      ],
+      [
+        "nvidia-cuda-crt",
+        "13.4.92"
+      ],
+      [
+        "nvidia-cuda-cupti",
+        "13.0.85"
+      ],
+      [
+        "nvidia-cuda-nvcc",
+        "13.4.92"
+      ],
+      [
+        "nvidia-cuda-nvdisasm",
+        "13.4.92"
+      ],
+      [
+        "nvidia-cuda-nvrtc",
+        "13.0.88"
+      ],
+      [
+        "nvidia-cuda-runtime",
+        "13.0.96"
+      ],
+      [
+        "nvidia-cudnn-cu13",
+        "9.20.0.48"
+      ],
+      [
+        "nvidia-cudnn-frontend",
+        "1.30.0"
+      ],
+      [
+        "nvidia-cufft",
+        "12.0.0.61"
+      ],
+      [
+        "nvidia-cufile",
+        "1.15.1.6"
+      ],
+      [
+        "nvidia-curand",
+        "10.4.0.35"
+      ],
+      [
+        "nvidia-cusolver",
+        "12.0.4.66"
+      ],
+      [
+        "nvidia-cusparse",
+        "12.6.3.3"
+      ],
+      [
+        "nvidia-cusparselt-cu13",
+        "0.8.1"
+      ],
+      [
+        "nvidia-cutlass-dsl",
+        "4.7.1"
+      ],
+      [
+        "nvidia-cutlass-dsl-libs-base",
+        "4.7.1"
+      ],
+      [
+        "nvidia-cutlass-dsl-libs-core",
+        "4.7.1"
+      ],
+      [
+        "nvidia-cutlass-dsl-libs-cu12",
+        "4.7.1"
+      ],
+      [
+        "nvidia-cutlass-dsl-libs-cu13",
+        "4.7.1"
+      ],
+      [
+        "nvidia-ml-py",
+        "13.615.71"
+      ],
+      [
+        "nvidia-nccl-cu13",
+        "2.29.7"
+      ],
+      [
+        "nvidia-nvjitlink",
+        "13.4.92"
+      ],
+      [
+        "nvidia-nvshmem-cu13",
+        "3.4.5"
+      ],
+      [
+        "nvidia-nvtx",
+        "13.0.85"
+      ],
+      [
+        "nvidia-nvvm",
+        "13.4.92"
+      ],
+      [
+        "nvtx",
+        "0.2.15"
+      ],
+      [
+        "openai",
+        "2.54.0"
+      ],
+      [
+        "openai-harmony",
+        "0.0.8"
+      ],
+      [
+        "opencensus",
+        "0.11.4"
+      ],
+      [
+        "opencensus-context",
+        "0.1.3"
+      ],
+      [
+        "opencv-python-headless",
+        "5.0.0.93"
+      ],
+      [
+        "opentelemetry-api",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-exporter-http-transport",
+        "0.66b1"
+      ],
+      [
+        "opentelemetry-exporter-otlp",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-exporter-otlp-common",
+        "0.66b1"
+      ],
+      [
+        "opentelemetry-exporter-otlp-proto-common",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-exporter-otlp-proto-grpc",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-exporter-otlp-proto-http",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-exporter-prometheus",
+        "0.66b1"
+      ],
+      [
+        "opentelemetry-proto",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-sdk",
+        "1.45.1"
+      ],
+      [
+        "opentelemetry-semantic-conventions",
+        "0.66b1"
+      ],
+      [
+        "opentelemetry-semantic-conventions-ai",
+        "0.5.1"
+      ],
+      [
+        "outlines_core",
+        "0.2.14"
+      ],
+      [
+        "packaging",
+        "26.3"
+      ],
+      [
+        "partial-json-parser",
+        "0.2.1.1.post7"
+      ],
+      [
+        "pathlib_abc",
+        "0.5.2"
+      ],
+      [
+        "pathspec",
+        "1.1.1"
+      ],
+      [
+        "pillow",
+        "12.3.0"
+      ],
+      [
+        "pip",
+        "25.0.1"
+      ],
+      [
+        "platformdirs",
+        "4.12.3"
+      ],
+      [
+        "pluggy",
+        "1.6.0"
+      ],
+      [
+        "postgrest",
+        "2.32.0"
+      ],
+      [
+        "prometheus-fastapi-instrumentator",
+        "8.1.0"
+      ],
+      [
+        "prometheus_client",
+        "0.26.0"
+      ],
+      [
+        "propcache",
+        "0.5.4"
+      ],
+      [
+        "proto-plus",
+        "1.29.0"
+      ],
+      [
+        "protobuf",
+        "7.36.2"
+      ],
+      [
+        "psutil",
+        "7.2.2"
+      ],
+      [
+        "py-cpuinfo",
+        "9.0.0"
+      ],
+      [
+        "py-spy",
+        "0.4.2"
+      ],
+      [
+        "pyarrow",
+        "25.0.1"
+      ],
+      [
+        "pyasn1",
+        "0.6.4"
+      ],
+      [
+        "pyasn1_modules",
+        "0.4.2"
+      ],
+      [
+        "pybase64",
+        "1.5.1"
+      ],
+      [
+        "pycountry",
+        "26.2.16"
+      ],
+      [
+        "pycparser",
+        "3.0"
+      ],
+      [
+        "pydantic",
+        "2.13.5"
+      ],
+      [
+        "pydantic-extra-types",
+        "2.11.1"
+      ],
+      [
+        "pydantic-settings",
+        "2.15.0"
+      ],
+      [
+        "pydantic_core",
+        "2.46.5"
+      ],
+      [
+        "pynvvideocodec",
+        "2.0.4"
+      ],
+      [
+        "pyparsing",
+        "3.3.3"
+      ],
+      [
+        "pytest",
+        "9.1.1"
+      ],
+      [
+        "python-dateutil",
+        "2.9.0.post0"
+      ],
+      [
+        "python-discovery",
+        "1.6.1"
+      ],
+      [
+        "python-dotenv",
+        "1.2.4"
+      ],
+      [
+        "python-json-logger",
+        "4.2.0"
+      ],
+      [
+        "python-multipart",
+        "0.0.32"
+      ],
+      [
+        "pyzmq",
+        "27.2.0"
+      ],
+      [
+        "quack-kernels",
+        "0.6.5"
+      ],
+      [
+        "ray",
+        "2.55.1"
+      ],
+      [
+        "realtime",
+        "2.32.0"
+      ],
+      [
+        "referencing",
+        "0.37.0"
+      ],
+      [
+        "regex",
+        "2026.9.29"
+      ],
+      [
+        "requests",
+        "2.34.2"
+      ],
+      [
+        "rich",
+        "15.0.0"
+      ],
+      [
+        "rich-toolkit",
+        "0.20.5"
+      ],
+      [
+        "rignore",
+        "0.8.1"
+      ],
+      [
+        "rpds-py",
+        "2026.9.1"
+      ],
+      [
+        "s3transfer",
+        "0.19.2"
+      ],
+      [
+        "safetensors",
+        "0.8.0"
+      ],
+      [
+        "scantree",
+        "0.0.4"
+      ],
+      [
+        "sentencepiece",
+        "0.2.2"
+      ],
+      [
+        "sentry-sdk",
+        "2.71.0"
+      ],
+      [
+        "setproctitle",
+        "1.3.8"
+      ],
+      [
+        "setuptools",
+        "80.10.2"
+      ],
+      [
+        "shellingham",
+        "1.5.4"
+      ],
+      [
+        "shortuuid",
+        "1.0.13"
+      ],
+      [
+        "six",
+        "1.17.0"
+      ],
+      [
+        "smart_open",
+        "8.0.2"
+      ],
+      [
+        "sniffio",
+        "1.3.1"
+      ],
+      [
+        "sse-starlette",
+        "3.5.0"
+      ],
+      [
+        "starlette",
+        "1.7.0"
+      ],
+      [
+        "storage3",
+        "2.32.0"
+      ],
+      [
+        "supabase",
+        "2.32.0"
+      ],
+      [
+        "supabase-auth",
+        "2.32.0"
+      ],
+      [
+        "supabase-functions",
+        "2.32.0"
+      ],
+      [
+        "supervisor",
+        "4.3.0"
+      ],
+      [
+        "sympy",
+        "1.14.0"
+      ],
+      [
+        "tabulate",
+        "0.10.0"
+      ],
+      [
+        "tenacity",
+        "9.1.4"
+      ],
+      [
+        "tiktoken",
+        "0.14.0"
+      ],
+      [
+        "tilelang",
+        "0.1.12"
+      ],
+      [
+        "tokenizers",
+        "0.23.2"
+      ],
+      [
+        "tokenspeed-mla",
+        "0.1.8"
+      ],
+      [
+        "tokenspeed-triton",
+        "3.8.10.post20260920"
+      ],
+      [
+        "toml",
+        "0.10.2"
+      ],
+      [
+        "torch",
+        "2.13.0"
+      ],
+      [
+        "torch_c_dlpack_ext",
+        "0.1.5"
+      ],
+      [
+        "torchaudio",
+        "2.11.0"
+      ],
+      [
+        "torchcodec",
+        "0.17.0"
+      ],
+      [
+        "torchvision",
+        "0.28.0"
+      ],
+      [
+        "tqdm",
+        "4.70.1"
+      ],
+      [
+        "transformers",
+        "5.19.0"
+      ],
+      [
+        "triton",
+        "3.7.1"
+      ],
+      [
+        "truststore",
+        "0.10.4"
+      ],
+      [
+        "typer",
+        "0.27.3"
+      ],
+      [
+        "typing-inspection",
+        "0.4.4"
+      ],
+      [
+        "typing_extensions",
+        "4.16.0"
+      ],
+      [
+        "universal_pathlib",
+        "0.3.10"
+      ],
+      [
+        "urllib3",
+        "2.8.0"
+      ],
+      [
+        "uvicorn",
+        "0.54.0"
+      ],
+      [
+        "uvloop",
+        "0.23.0"
+      ],
+      [
+        "virtualenv",
+        "21.14.5"
+      ],
+      [
+        "vllm",
+        "0.30.0"
+      ],
+      [
+        "watchfiles",
+        "1.3.0"
+      ],
+      [
+        "websockets",
+        "15.0.1"
+      ],
+      [
+        "wrapt",
+        "2.5.0"
+      ],
+      [
+        "xgrammar",
+        "0.2.8"
+      ],
+      [
+        "yarl",
+        "1.25.1"
+      ],
+      [
+        "z3-solver",
+        "4.15.4.0"
+      ],
+      [
+        "zipp",
+        "4.1.1"
+      ]
+    ]
+  },
+  "trajectory_input_sha256": null,
+  "expected_outcomes_sha256": null,
+  "upstream": {
+    "terminal-bench": {
+      "url": "https://github.com/harbor-framework/terminal-bench.git",
+      "commit": "1dcda8716784493721921c23e4bc7f7d988b4494",
+      "sparse": [
+        "scripts",
+        "docs",
+        ".github"
+      ]
+    },
+    "harbor-validation": {
+      "url": "https://github.com/marin-community/harbor.git",
+      "sparse": [
+        "src",
+        "packages"
+      ],
+      "commit": "7faf878c14b6d72737579ec936ebf5f74ba3194d"
+    },
+    "harden-v0": {
+      "url": "https://github.com/few-sh/harden-v0.git",
+      "commit": "342b8474e0c0cf96e4a8313fd2e26c7a11d51193"
+    }
+  },
+  "implementation": {
+    "config/clusters.py": "bc2d58a7ffbe63948bda15e8748059c750d33bf2d1e99c50173dd9a9b6e59e6d",
+    "config/models.py": "2eec0a437188c9eb08cc732af4f57cd63ffb9616bde4f27379edb0a287a5820a",
+    "config/runtime.py": "7429db6f2620c0b25f229ddee62a2ca0d7f82950e835a8e7b05ed50f5786fe70",
+    "data/annotate_dataset/capability_taxonomy.txt": "10d929a31dde33f96858fbd844afc04bc73a40b858dcf5602f48e868d7cd0351",
+    "data/annotate_dataset/domain_taxonomy.txt": "ad76b324b05270577f8735c426c53de942fa2af0e448622673bc35349cea7802",
+    "data/annotate_dataset/evaluation_suite.txt": "a6aa12c845a9827268059311fa841a740950c10b3aa341b22cd6f331d527d956",
+    "data/annotate_dataset/prompt_template.txt": "1db5eddfe6b07e3bf9382f63cd96f8c3bd35ffb30446737589b72592c835884e",
+    "data/seta/patch.py": "9a2f75a5e59c488a3405e0003e6d70bf9f40a4810bca1cb267f1bd55093a2d19",
+    "data/utils/__init__.py": "354a01fbe0f2fd0481839ec8e08c99d1205b6fcc6c24d7e7bdb846ee6df34865",
+    "data/utils/cli.py": "5f7cc15c9652822a42fb17c680cb2ee8492f4e59506cf886b7e2cb2f28f4bd4a",
+    "data/utils/environment_counts.py": "05e8bc86a4c7b5f0a0f9a22485846588de0b9f640dc0a8a954f0fa817de8dbc0",
+    "data/utils/fineenvs/prepare_inputs.py": "e2ecd0a5dcce3a4e9adc71dddcca9738cd497af2d6f91120a25900fae463fbaf",
+    "data/utils/fineenvs/stage_inputs.py": "0e2d0e469d2870a41fd3e8b7e2d21afd8d4f6bd96048fa581dbfa3cca20c8e23",
+    "data/utils/full_source/__init__.py": "d9c233f681a11a99f34d1bed30837b035649fdf800f166c0ca289361a6f9830d",
+    "data/utils/full_source/convert_prime.py": "5501a58f1c5169981f89196208c51a9906cd5a8c41e6a53972c0cc7742eee47a",
+    "data/utils/full_source/copy_source.py": "f416d12e58878e7cec77411d7075a04095a3586870a4c8253f6048b310d2848b",
+    "data/utils/full_source/fetch_prime.py": "7a37868e5e1d45d1ccab2cc6f98b2df34838da7d0021e89df24d196a0316ab09",
+    "data/utils/full_source/harbor_parquet.py": "e0b83fdc5f41bb70267fa8cb23fc040a97372e01db0de90bfc99835d3b59348d",
+    "data/utils/full_source/multiswe_grade.py": "49d15d9da18badcee84661aad59a963fbf2e3431d3c2abcb155f5fbf330fea6c",
+    "data/utils/full_source/prepare_multiswe.py": "709d60ba9ecb05c327b22e70b9a530637c2625bb9f28a77f9d1c3416904e9a9b",
+    "data/utils/full_source/prepare_scale.py": "714e87ec612f2f7cc4304941ef4e35b9c893a874dabc1c0739b36a11ead12fd0",
+    "data/utils/full_source/prepare_swelego.py": "763d1f0aeb0555e83222b1b498ae1cd73d1708366ca150d5ca21e8b6278d579f",
+    "data/utils/full_source/swelego_grade.py": "fede48517f440983dc8d0c5dc15be848097f749a21150b2265480d48c3f42264",
+    "data/utils/historical_release.py": "80b99b934e602c13dac5dcf27a87c2f5fd71d5593d495c5074751a757fa18b45",
+    "data/utils/instruction_loop.py": "ac8a2bc9b62e3c28fd6bfc9ba2ad85734df38e9e4260a0bbad51549f49e9224f",
+    "data/utils/instruction_resume.py": "1bdd4b92ca1671d98e40a78071c2e65211cad43ec04e87dc256867dc8320eaa2",
+    "data/utils/maven_proxy.py": "b0468122f9dc071b74862a3f17a267f57d374996c32f27eabb5927439e49f0ad",
+    "data/utils/pack_git_harbor_source.py": "fb28230c8bb2afae5d3b672d9c77ae6e2606d22f4a4853b12967f99adbe0272d",
+    "data/utils/patch_reporting.py": "927dc5593f0556b307e2eb70256843cbb712bd7abe89b6646df72fdcf3155afd",
+    "data/utils/prepare_hf_tree_pilot.py": "203d0626674bcbd318e0de532f2688fbd1eaf23bd0ef0cd6626e22f54bcca826",
+    "data/utils/prepare_tasktrove_pilots.py": "944b152321be6ea2ea2c7b833bd838209221be0932a03b8c8e4a58bfa0ba49c3",
+    "data/utils/resolve_absolute_paths.py": "c0780aee2bb048ed61f84934a1f61b9f5b8505b636ba788399d1961929d5a1de",
+    "data/utils/resolve_pip_pins.py": "e21615ace6502675880b89888d22773303ebc6347ad761edb3c3dad864974b20",
+    "data/utils/task_archive.py": "01ab5693707f6cd2667b42a3e6b6b901d2f8ca4727d199e267a3393522dcb181",
+    "harbor_patches/artifact_transfer.py": "90fe9ff27a2ca707ed151b84da8221b4958e67bf691e214364eb8924a8cf3231",
+    "harbor_patches/bridge_client.py": "c9873876e5975b7c9f234fe0ee1da9296e3536296df9adeb8a71f2340c156b8a",
+    "harbor_patches/bridge_server.py": "319f2b965ed12a5b34b832da68eb99429176ffb6955ad52d7b95746ab6ef9582",
+    "harbor_patches/bridge_timeouts.py": "1078b7c60ecd2c07cedf2dd3e3130524d70e6881558f3a3ac9fe9d0c35fb6aab",
+    "harbor_patches/bridge_worker.py": "939d11d6ef1b73bd74ca1ad951901eeff4fea3f04164ac0561676f0e1181c90b",
+    "harbor_patches/command_cancellation.py": "909c862aafd0e51a63acd51f9ecdbc5c30d41db264aada761bc98496ef4d8795",
+    "harbor_patches/container_env.py": "32152c1fb910f211742cc850fb13ede30443f709b0917d00577a60b944997a60",
+    "harbor_patches/docker_path.py": "44191673909a35f302be95a3bceb5f1d9024557558aad03a715ed273cd9c1e73",
+    "harbor_patches/exec_environment.py": "cecab5bbe66944a67d6b954195ac3f2e0673c24efee23c906dd9fa1084e79479",
+    "harbor_patches/fakeroot_ipc.py": "45c29a3f86db5a9f1a789f3d359a2dd942c278b71cd8b38c77d0cee7c2415fee",
+    "harbor_patches/image_build.py": "aac9f11a631cb073f3dd2eea913ab62b12b3d75c911358fb91cabb2faa7dcf64",
+    "harbor_patches/image_context.py": "70fc8bab6e6db0ae5a23a0eaba83c023ceb926852fd6a5695c9ba94115291c36",
+    "harbor_patches/local_image_base.py": "bfef0cabe64538dec751076d492bcebfe8c8b0f2fed931fda816896b0722e3d9",
+    "harbor_patches/oom_recovery.py": "b60e1a999260a2d0dc3d4a0674d0a2c2d88bd1605678a3b820e7f9850cb25703",
+    "harbor_patches/ot_execution/bin/ctest": "e6a56a7aee35b22d2af45f0cb91d42a0973b6ae236b52ec228037a5c3d1a6687",
+    "harbor_patches/ot_execution/bin/go": "80e306f655058422f872e4fe4f85e0eced2ac549b8234c1a63b79f1074d29be1",
+    "harbor_patches/ot_execution/bin/mvn": "81aeb960adc4b7a91a008d8d78c056f87e53e269b15f336939d331267963a60d",
+    "harbor_patches/ot_execution/gosrc/go.mod": "d3d6a3903112807cf45d0cf756f1879da4695d57d79190d6c262263cdfd65d80",
+    "harbor_patches/ot_execution/gosrc/main.go": "7be08795b7cd810beabf1ddc2632b740e3b42e2d66e8221a375dfadc2ac68fc8",
+    "harbor_patches/ot_execution/junit_reports.sh": "37a1db6a8ae743df09d044c93fb8237f75809a9c0573a1f4b4b8c31e45212621",
+    "harbor_patches/ot_execution/ot_node_execution.js": "bcd891f89d7cc1f4a2fac9ed318699488933dcc8ff0841a94a1cf586cacb0616",
+    "harbor_patches/ot_execution/profile.sh": "17aa854f39d7c01b5cb4b6d483f41a813f363724d3201aa0b1fc28abdd200ee1",
+    "harbor_patches/ot_pytest_execution.py": "fbb36eba9486a7672862d1203ee24748ab82b716d9c0fb6cdc9e2b5e3cbef41e",
+    "harbor_patches/output_recovery.py": "25ea50c541a56186903353d037bfe970dcec4011e4c6ad94872bacb7da370eb7",
+    "harbor_patches/protected_step.py": "6d96b344fb93e0fc3d102bd1d86855d4547a3990730b4e68da4b8da03bbed48a",
+    "harbor_patches/reasoning_field.py": "1cc4a0a617e694c1d80512f904cf8c4f740e9bbbe8473deb12f535869ab5ca82",
+    "harbor_patches/resource_usage.py": "30b2e1c9c8826d78bc2fe126c5eca2be932b3f02a5f40dd179b88868a8ae883c",
+    "harbor_patches/sitecustomize.py": "75866b7e84fa80744cab504bb946965c796b8c45984006a9268300005e5bba32",
+    "harbor_patches/start_wait.py": "1a333ec810c438c633cfaf7968a77b982cb47095aa29f764b38319ae292b9089",
+    "harbor_patches/tmux_capture.py": "6c5346fd75d1c4a460b6ec0af272ed3cfe57dcf8204dd02a7d02a5b1ca8dfc68",
+    "harbor_patches/tmux_diagnostics.py": "41db4d89f04587e7485db77246e0469e097cbdf1cb5f37f7f442b95ae099018b",
+    "harbor_patches/tmux_runtime.py": "0de81026d5b5b8308f2e7186fec1760198560282a29ce8a463c1665c11f71088",
+    "harbor_patches/tmux_socket.py": "d22f682ae28f7c0bd093e82a300c3de1b4a2e41753c0adc270f5744e794304e4",
+    "harbor_patches/trial_step.py": "8fee42ea134d78bdf8527aa99657b4336fbab69e0cf5a31ea5cf33e9795b5499",
+    "harbor_patches/upload_runtime.py": "02965c6ce9319014732849601f544893dc37411d7457d4177a899f5532fda2cf",
+    "harbor_patches/verifier_setup.py": "97be873552fe710949ad3fe1825a9d083a78523bc69d1b1dd036604a7271baf9",
+    "harbor_patches/workdir_seed.py": "a9744ba99aa5bb210cc1aa05ec30f6e22656fad13b38095f35a5c4613436a5e8",
+    "hpc/helma/proxy.sh": "ab27bb020a914f87e7ceaa592c00d505828e9808a5dbf33231421bbf392e9c5b",
+    "hpc/helma/validation.sbatch": "87dcd6542aa24d6026f63f01db96041343b72afda5be36c2ac7a7b5093a93d32",
+    "hpc/image_cache.py": "f9700584d3f042d08aa7f9129ab9c16dc4183350e216e50e7ddb1230c6e5a2c2",
+    "hpc/local_assets.py": "0da6c6e83d0d3c232b3d2c6d2a871c169c2dde18fc4c5801905ddc5667ac4653",
+    "hpc/local_runtime.py": "cafc4f0a63d550bf78a640aecf2f24350611ff5f76e251041575a28c2a89638d",
+    "hpc/model_startup.py": "1e26eb9be4fb8ac43ba288530ab066651e2f82ae5503cdd138b7f4a20849be62",
+    "hpc/rwth/validation.sbatch": "2979f7faf7b12c5097c94e2eb94389fabcafc45fbe4ed44ded0754a4e3ce2828",
+    "hpc/validation.sh": "96355b9b69e8f694751099c33b676b00eed4e4de8eddf5f68e3ec193300f9ae2",
+    "hpc/validation_submit.py": "0dd96a0cc829941a6e14ecec4916afe55cc8c280b8b103addc4684a7253a58cc",
+    "hpc/validation_worker.py": "e0c3965ce682d80dcf7b97c6c6b569f81d2b7e33e28be90f5567024bb155ef36",
+    "hpc/zih/validation.sbatch": "ac1757354d20d19170ba3eb1a1561f1122fd453a9b5f23f9c64172ca98f8fb2c",
+    "requirements.txt": "fb46a4637021c792e0a01bec69d66660ed13cd71ceab444503dcd86b6a098ebc",
+    "validation/__init__.py": "949bf1a249cb47e41e65649a951552a6d0633c9677da4b2f5e12a0da0368f075",
+    "validation/checkpoints/__init__.py": "47324c978da5e5f36bafaedb1d442f51f428ea63e0a962b6545e399f412b4fbe",
+    "validation/checkpoints/recover_checkpoint.py": "b102b9c3aab4d1e4f50a0a35fd14817d2e4ff657f75471b2b01e1d005ccb3910",
+    "validation/checkpoints/stage3.py": "06a4ca39c90875f27e83dc97405b05c526dc4c03d1208eb27b1b3abe1c028c83",
+    "validation/checkpoints/static_resume.py": "b45c62bcae440fd827e3e3f50c93334dde6aa6f4458719da3a0d49dd73e7fe66",
+    "validation/checks/__init__.py": "1cf3dfe0db636a3c6966e04caac113b65a98077b8b5539bd3c6f52d5f0e4d640",
+    "validation/checks/agent_run.py": "871762885edd187839045d84b577184a38a2e81ffdffdc1fad94c075a2c0df3e",
+    "validation/checks/check_images.py": "0f0e50911ac46c5e076d4e771d8a689bcf130bbe1305e44dccdd6daaf52e733c",
+    "validation/checks/check_isolation.py": "46c1c39438258e0a71f0c487ff2171b0bd979d6090fb2d3cecad4eb2563d7cd4",
+    "validation/checks/check_reproducible.py": "3f915577b3b2880f5d70ee894ba7c06ec605222d11ce084b92326b158d607d8f",
+    "validation/checks/check_terminal_bench.py": "9f8f1f2fea2adfe68b46aeb7984a63f37c0a7d4ca870c1586b11f3175ba87ee1",
+    "validation/checks/command_metrics.py": "16f2242cb6c034e5d36b3617080d50beb0d7ed22c91c7207ad13c3568fb1d07a",
+    "validation/checks/dataset_checks.py": "ae9790971f969959c4168c0a693fd5fe8eea5453030716e0776cfca5534642ac",
+    "validation/checks/environment.py": "776e86df406c253ad7850d0b358563fece205cfd8c8ea91399fef593e5d6aee7",
+    "validation/checks/fineenvs_verifier.py": "318d4111c113e8731b6b59ebaca79943b1fbea1f4c0e01c0cd73cd1b1c90f3db",
+    "validation/checks/instruction_paths.py": "d61851018b58a4c153c720a9eab3ad0520390efda37a5f47381fc27abab4f559",
+    "validation/checks/instruction_suffix.py": "fa9722069acc79b063079b14345b082155de6fc0cedeabdd4ecee68f1beb453d",
+    "validation/checks/passk_completion.py": "9fadfea4f1a72573b5421603db468d62a51810dca4bdadb94d0ee45efaa69f0d",
+    "validation/checks/passk_rewards.py": "19807eba9e3d7b3c442f43c1508238a5d772ec2132def8268e42f957d2e445ef",
+    "validation/checks/path_cache.py": "0d692382cfc6d218acfc48ef927633ced395926f8315ae78504dc63204971743",
+    "validation/checks/reference_files.py": "ddb1f97d0f684c5789811c65695926dc4972c488df36ec58cf2cc55c1ee80759",
+    "validation/checks/repository_repair.py": "d90cb7fcf40dfa423d503f1b27d2d1640b5a00f3f993b6e7e0e9b67949087b46",
+    "validation/checks/review_results.py": "c22457df68dcdfa6ac24e56055612bb3886a6a251feb0393d934b3c6ddbb814e",
+    "validation/checks/reviewed_nop_zeros.py": "bbda63078c26b6f52af8005c0de292d5ca6acdaac4d8291978104d0513350346",
+    "validation/checks/reward_metrics.py": "e1bd981f2842802aef77c8e56105ae594649bad89bb7514b310f11b5d6ec694f",
+    "validation/checks/supplied_references.py": "fb6718b840d25dc415267e54100477bdad51eb544f2b9c20fadef24813039850",
+    "validation/checks/task_toml_schema.py": "f7f80df455cf6f291b0284b8d9230c22ab1fabcca10b8baed13cbb4a21cfb725",
+    "validation/checks/trace_metrics.py": "ffa2ef63f6c84b193cd50503ece64f6819abc82b894c041ae9c7fb601ef33fe1",
+    "validation/checks/trace_outcomes.py": "570e8bdb6930588cb87b9f3728518cae549fd20da595808a2ecef7fed75dddc0",
+    "validation/contract.py": "3e23c7904162112ff2cc6516ba57b20120d4d40cfc33aabfc9b8b124cb78a5c8",
+    "validation/data/__init__.py": "aea9b5b7003cefb8631b11ac533ee2c3de8fe332ae56abbbe391bf573461ece4",
+    "validation/data/create_pilot.py": "384e8811ebf8227d296a240ca72904b0eea82ee4a938de8c96e58d6a8e870643",
+    "validation/data/materialize.py": "e2608a71b86249839de3d79ed37c6f8d26bcfca873c392461c8a711ec9c3d3ca",
+    "validation/data/select_smoke_sample.py": "739c48d00b026411c60da1319d6a3a2397e1994c2fcb69f2b806eebd6eb7bd3d",
+    "validation/data/selection.py": "66d77c1d88c7a0ecb2d1233624c98f60b15e7d443f027ef9fa102bfa9558f94b",
+    "validation/expected_outcomes.py": "d0d8e96a6fc2407e1af72dff6acb7d90ddc2313597bb4faf84a0e894123391ef",
+    "validation/merge_reports.py": "9ac3b267b51c09de718a823ac9f78cc55cafcbfe1f0471d8e3018486aafd5f47",
+    "validation/publishing/__init__.py": "b70229ff19247adf57b8666b20765c17a40dd222534aca331c10ef7c3fbaa35f",
+    "validation/publishing/annotation.py": "d8a345f1c4b654b66d6b1d55641dec0eda13e6e46625fd84761c706c39eeb89b",
+    "validation/publishing/annotation_runtime.py": "44995b0f9a255a4bc594873a1a07859c53af36a090df352c22f4ba63e3b0e551",
+    "validation/publishing/combine.py": "01bf722c0d61a7dacf82ea04b855fc51622ed420a65045d22010a3a5cce7c944",
+    "validation/publishing/evaluation_settings.py": "2223e2134a25a33feaf8272f1aa1e883aeb0a5498971229373c99cd8661d4b26",
+    "validation/publishing/image_release.py": "b1688ff32bfea4af25c5f7f19045b06ca6611bbfd2f336941d75f455348685fa",
+    "validation/publishing/outcome_labels.py": "c359ef33710650763f5f09e86a02b17a0402c24096890c1c2e1fc6fc110cef58",
+    "validation/publishing/parquet.py": "9f6b8c76ea3a5fa8cf821c1d03e9b1f2e623034f6fd4bddba86af61f1dbe83bc",
+    "validation/publishing/patch_provenance.py": "447016031c28c362098817de5bda0f0fe11426d70ca9014b6028cb75c2803ef5",
+    "validation/publishing/pr_runtime.py": "ce945eef7cfa4c3e0f458a412a215f8eb311f25da6fed3cf6bb265134f7a8fd0",
+    "validation/publishing/pr_tables.py": "bd9d9c340abda94477c5b9cd720fa3e3a931dd771c0290adda0c0ecddcd1bd6a",
+    "validation/publishing/publish.py": "27f8acc748625e03ee1f1bb23d0d932a8a00d29ab00444d3fb2b73cd809addd5",
+    "validation/publishing/publish_pass_counts.py": "ec6cca72b5c0ace54f7383bbc0bc293f16bf43ec9198171bc6537c67a85a75f6",
+    "validation/publishing/trace_budget_subset.py": "42dfd7b1136f28f5d3967e4db11f4904cfd81e8d50536bb8867a976ecf9ef680",
+    "validation/publishing/trace_outcome_stats.py": "d22537bbe914bb55aa1ec24f6bb6010ec215284819faff5aaa0159b5a0a3154a",
+    "validation/publishing/trace_stats.py": "a950fbeebf48cb9b4716629ffebcd98476a7e320a6b72e1eb9eef4ec16483c5e",
+    "validation/publishing/trajectories.py": "97eb58ad87160d7e7fac36033f467c1f16afadeb0bc94a2ede2e7db09ce1d511",
+    "validation/reporting/__init__.py": "047952e98c632807db2e5f31fb90351672f8cb20f4dca43edcffdfd9c504b08a",
+    "validation/reporting/plot_results.py": "3e3bdc00f8b2100de4f96bdb48ce6fd87eff518a6bb380f9b07c3da352ad7db6",
+    "validation/reporting/report.py": "0a2ec798033cf85186d518b028f5a2d7a16bd2952ae46c130c63a844d6f9fab6",
+    "validation/reporting/review_pricing.py": "31c1acadb9923cc6f1144647c3784fa20538b2a6ffe14b1dc279236e062d40ae",
+    "validation/reporting/review_usage.py": "d03f30241f2017365ef578a5dab2f0e857116d0f25969bec7566c42ef8c2b271",
+    "validation/rubrics/__init__.py": "2225b59b999eb8a99cdcb8616bdf6b6c6553fd7f5094ee415e883cb8dbf9437d",
+    "validation/rubrics/analysis.py": "8e4b57a8e7d549a14d8d091f304cf5976956a80bdd506b7979a8effa179d083a",
+    "validation/rubrics/labeled_examples/charlie_hackable/run_reviews.py": "f46f7b53c8804daedfcd5ce6449f81568a341051c95e9dc8a03f9cb615f7d9b4",
+    "validation/rubrics/labeled_examples/tb3_tb4_assumed_positive/scripts/fetch_samples.py": "c5c78365099d883a6c51020ffe644ab5999f08d86007bcc7663666ba2c0da2cd",
+    "validation/rubrics/labeled_examples/tb3_tb4_assumed_positive/scripts/summarize_reviews.py": "ff3de1243b2add5ab5568e6dbb3a792822d5b730a2884b50109d96d11155a6f8",
+    "validation/rubrics/labeled_examples/tb_human_labeled/scripts/compare_reviews.py": "30f85244cd8be8e56c3556d2cf46e30865160c6aeb9dd3bcfaff91d0fb504c1f",
+    "validation/rubrics/labeled_examples/tb_human_labeled/scripts/fetch_tasks.py": "6e2ed4da122ae3fb3023ee22525b31c2638695e1a379022e5addaad15fc59ba7",
+    "validation/rubrics/labeled_examples/tb_human_labeled/scripts/map_custom_labels.py": "f9a68ab9f9c5eb1aa535d528c762049542fefedd345cd7028eaa5bc66fc87728",
+    "validation/rubrics/labeled_examples/tb_human_labeled/scripts/resolve_tasks.py": "01847613e354bff7832cf2d42faadfa7cbb37cc43468764d8d46591c1aaffb97",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0001/tests/test_state.py": "9c684396c092879e8206df42994bc24c8bba3e45ea29d7465ed281ed5377f460",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0002/tests/test_state.py": "a9e6499e78078132be65262899e1a0f897c03f0bb7415dcacb4c7416783397bf",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0003/tests/test_state.py": "e3f220f193b9db1feef89cf465589e4c1da4dc1e19d02eb09301e30dbf8336bc",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0004/tests/test_state.py": "ab1d6f6eeb24f083fa8f520a536b41091fa158c26376e334ebaf3b8f95895862",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0005/tests/test_state.py": "08eb89a8c672572ae83937b58044c39b67d3c6b552da99606246dc53d7741621",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0006/tests/test_state.py": "4ae75cfbf94554000c01a2d2b65bd5a8f394f12a56b96e7bf29830449f744585",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0007/tests/test_state.py": "4f6aa3decc2444b36834588cff089496acc5354d80bff9aad451d350514eeaf5",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0008/tests/test_state.py": "4f6aa3decc2444b36834588cff089496acc5354d80bff9aad451d350514eeaf5",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0009/tests/test_state.py": "bf7e901076c949c2a7e36da2fd921d000e395e3e6730a98ff963e2c0d98faefe",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0010/tests/test_state.py": "5b9b2db1509f0d2da37ee162b0cdbe77d9ce7b2da205192d4a296352a68c07a2",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0011/tests/test_state.py": "a7eb245defaaddd97c7e31687195ea86958d93073df9f5122d8c0f06f0a4b347",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0012/tests/test_state.py": "0654fe4ffc3f25362a670db142af29adaf174ea7010718aceb7e342adbebb1d3",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0013/tests/test_state.py": "e7c46f82c44d997a6315b1ef833fd288fc2d9515ee95a4352abda301a47af922",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0014/tests/test_state.py": "83dae6ba77606ee6142f94069a9d8035218c6e9ab9b10039158051a98aa25f17",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0015/tests/test_state.py": "2a469a878055d3f14e91f505e94e3292d97bfc08190cd85af72268251c53c12c",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0016/tests/test_state.py": "bc2faf07f2722a00a803b34b6d4bcfa82648bdaf1ae1f19d19edbd4afc902dfd",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0017/tests/test_state.py": "19fd4eec5b36a67b34e52f79f6e30a718b3420f2291105a67d8f5e203c401e52",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0018/tests/test_state.py": "edec350130a882792b37026db73671d22cd1ec3a45eab3fd71c77b8a0d17bf38",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0019/tests/test_state.py": "c0016867914c659f20bf21fefc629da106396faa2d0e1f1a0ebb4bb164aac4c3",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0020/tests/test_state.py": "a721d5aa0f0d92fc81b29e2914e988040863ff9cac947a21bd6df7cceb19c483",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0021/tests/test_state.py": "8383d142d88c42937f9f88f51021a1e1b2fcc45bda808c0c55ed59fa682400fe",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0022/tests/test_outputs.py": "75442afc1bf002ea52874d68c2aadd6f9fd9afbfbc1e40a4055a233aa042f40a",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0023/tests/test_outputs.py": "0f7134c7dd17bfafe2efadde55ce6d0bbd0e3c2096aca825fb6080937c353cf0",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0024/tests/process.py": "3525f32100c994ee517ba06a355f6d2130c332857da017d921aa8d9252b7022d",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0024/tests/test_outputs.py": "57e4904177acac5807b51f7c32444bde1968ba0445855fa4333d9eb73865b59e",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0025/tests/test_outputs.py": "fbbb57a31ba922d610dcc9ef526b240cfc654c147dd9e16e7a13c4cdea389472",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0026/tests/test_outputs.py": "81a6e5e92c7351d215286438d2b3c343f44be09c2646efd0aab231e52f927d39",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0027/make.py": "46b95008df62fac7ca7fd7260baf36519de5718b91a4bde485c8d9b7d37f17c1",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0027/tests/test_outputs.py": "a540714e0156fc14564badedcb48d1f7b69a4b457d876036535439b30cb8f286",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0028/tests/test_outputs.py": "2dc14fc180734b6fbb0ee00f27c4f866071247b57a2217f579465ff8ed3bfc9e",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0029/task-deps/run.py": "9dd34b5ddac15b3db57b56ac37e61c1f352eb227dd24f22f967925a688694480",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0029/task-deps/test.py": "58e40482e67a4926117c6ebedbd4b559119ea026ebb2d5128f87c37b426ab61f",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0029/tests/test_outputs.py": "edaee4e792aeca299929d5a811ac386c302d7918fdb15bca4cb7043288a45de8",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0030/tests/test_outputs.py": "77511a789d27a2b0c13cc2529eb6a3967e6a17e87defc1250238a2284dd0ad60",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0031/climate_analyzer/analyze_climate.py": "112e26265c2b7a579403c9690f5d58cbc1bf59468bfcb7496e4fdbec9a565dd0",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0031/climate_analyzer/data_processor.py": "4bec5a0e8954fe69c04f248f5c4df3434fa1925f3d226518f3ffbad7bb8278e3",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0031/tests/test_outputs.py": "60348d5ee8a374c11851a4320d2e812d9498420c3d03152ef8e57cf393dabcd6",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0032/tests/test_outputs.py": "1de73eac00b0bba477599bbc0a393d5eea65e9c2d92fa73bec6e885aee543a4c",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0033/src/eval.py": "e0eb028480c98b65a0256f27bc22805d3b3fdbf4d6a58282b8dd2f648dac34a5",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0033/tests/test_outputs.py": "4755e83525412fce281370a665b6e5f4a552bd2f81b73f07f9fd8bbd3cd2ea10",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0034/tests/test_outputs.py": "ba1ad3e129c8096c9c0a2677a73dfa463a9c2808a2055b8f3fcd741a8530dc0c",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0035/tests/test_outputs.py": "822171f8efbb8c02f467c20c2aa4485cc63717bdac2f0bb06f1689d3e61c195b",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0036/label_documents.py": "7c2ff69f87765344c0a51f557273de01e7a5740fd600c37f490106cc982b4bf6",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0036/tests/test_outputs.py": "43fe21ae9e6a9ed1c57a90c6d33e6a4fab3b1e0d4acb40a55fb4a3cae8f52f21",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0037/tests/test_outputs.py": "cb8b06595253aa4020118c23a88536391312bf5cec41ef2479e28a740c725f42",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0038/reformat_data.py": "aeecff97fbee11c1fea50e3123bcf678a05326be0fa517545e6863a4c704b12f",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0038/tests/test_outputs.py": "ca645971f6b4ac2fc0b5ce83eac18df0bb936d7891ec73b1fcb6fc8d2fbe0884",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0039/tests/test_outputs.py": "e4e6b21fed6662adb4a42aba23d35609450392839a2282f3ba77c70c8d907c0b",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0040/tests/test_outputs.py": "a54b79c90c8a2e65618c517dd3fd8f51239d9615d1c5f57770a97c5e09100487",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0041/tests/test_outputs.py": "02df5d8f2ab15eb9ec39493c5ce57b8699853304cd8dd5b853e1dcb8c4a13327",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0042/tests/test_outputs.py": "18e8f2eedfcd64ee64d4163aa8398d6b8a6047b33115445a5cf3b44b7e9d941c",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0043/tests/test_outputs.py": "a7eb1197debe8a7083c750b160c42c5fd084686dde7df88aa86e70999c62b1c7",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0044/tests/test_outputs.py": "565a7617602f33c0f103d6c9b39c95e0cfe72ba8eadf4d3f859cabfd9235467e",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0045/tests/test_outputs.py": "045cc716c14efde3b0dcff5fc7c85ec5d18bfc6ce66f8b40a418fa2a3a4acda0",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0046/tests/test_outputs.py": "b165f8504bb93a901451b80c4ea73344b71b6af377ebdabef745fbcdd1d2993b",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0047/tests/test_state.py": "858650dc47bb32130186c6d050a1cb61971ad2234d7c52dae4a330b2edcf6c7a",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0048/tests/test_state.py": "e7353b18cda27ae814cd73479321877c31a21ea13fd9679812136c54bb7683ee",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0049/tests/test_outputs.py": "9d39800486b1b14f9b7053094caa37fab25194b78c289ad244d096ead8de8517",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0050/tests/test_outputs.py": "da31ead95b5fd905b945ee9203ce5ca716105a3c321d4eb8d83863f48504f384",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0051/climate_analyzer/analyze_climate.py": "112e26265c2b7a579403c9690f5d58cbc1bf59468bfcb7496e4fdbec9a565dd0",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0051/tests/test_outputs.py": "0c24195ff46527764c7f8ceda9c608a189ab4a01619282942fcd27a2ea02a977",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0052/tests/test_outputs.py": "bf90146ad901aee94d61c9655d821b71d19c3dd9d92d91c9fa08341bc823cdff",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0053/src/eval.py": "4dcf044f28a5656d3bffb35f81f8de9b7ad3dd15327d8e5f3dde65586beca827",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0053/tests/test_outputs.py": "a65b5a1264219a0a33ac4458d6f587ce41b25c19bd71bcd2c95ec337058a3e02",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0054/tests/test_outputs.py": "a77d75cbae86f951f2dcf17a7df11eabc8eecafa9af63a9aaf311d47d4891d13",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0055/tests/test_outputs.py": "822171f8efbb8c02f467c20c2aa4485cc63717bdac2f0bb06f1689d3e61c195b",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0056/tests/test_outputs.py": "e5774c67de61e37c5138e543079284435afa8c37d542e461d03b5566e4b92afb",
+    "validation/rubrics/labeled_examples/tb_human_labeled/tasks/task-0057/tests/test_outputs.py": "6144b70427d2605569607ee3ef69da5acc1253cb9dfd41fb19f67be53aaae952",
+    "validation/rubrics/run_review.py": "9ba8186cf268f5d18393614615b4ba5720f238bdc39cbb2abd1ca6e635a9b4c0",
+    "validation/rubrics/scripts/stage5_calibration.py": "ba6dbda3b9a596145980eb12c0ad3471b84abc35c58fc6758a6906c9ee72ebde",
+    "validation/run.py": "17874b254c5ac123fc528beea4833c25fdee141aeebe095fcce8091534539107",
+    "validation/shard_normalization.py": "f3cab1b2b05567528d20498bac899ab361e5b40dd00a446a065d0a90f4552bcf",
+    "validation/shards.py": "0c1cd8e9369f1bc094948182876512fd645620a33636e589e68eaecac3278a59",
+    "validation/stages/__init__.py": "8d31f873e5fa008a5e522a7c0af3873e0f291a8f0ae1745c7760eaceb160bf35",
+    "validation/stages/build_retries.py": "1cfaf813934a33e619a25be6209d91b5058894f74d003de2451fba1bf9600f1a",
+    "validation/stages/container_reuse.py": "54407f74fff9ebcbcb6835993d572a37537c41a35190fdb9ccb54889b5c68151",
+    "validation/stages/env_review.py": "a6acff791fc244e63afeb2afba1fc569e18fd64cfc9355e6bd33988930ff3f09",
+    "validation/stages/environment_preflight.py": "729cbd4cb547f5039f85c65d70fd529dbc6494a9a8a9faa98ab61d8783bad24f",
+    "validation/stages/execution_evidence.py": "ab9b5fc69558568c48431852a5e9b4360300e87d6f2a6196202d859026fed9d6",
+    "validation/stages/harbor.py": "36c0b2cee469fd7e15ac26807d919021558629ec0436842a971b94cc9da011c9",
+    "validation/stages/harden_entry.py": "a49fc58f18dd8fee9af91c239ccf293e15c27f0f3d14294e49865e84920fe746",
+    "validation/stages/image_inspector.py": "ec4dbdd73824d1ad7542ddbb3dd20593372b308248c7e81a5c44c207c634519f",
+    "validation/stages/image_review.py": "ecd05fb35ff34a2a97a8d75b4f9f94450b9e5b251656cfa0da34503bc4fe7d25",
+    "validation/stages/layout.py": "d1966f09641f078aa3ea99b8f6273b61bdab15848619047fc80d7e6d223da38f",
+    "validation/stages/normalize_paths.py": "fba86597c932e335ab204dbfc73470c5530bb240660c184cd062c1dd52abfb64",
+    "validation/stages/outcome_retries.py": "dcf437c82e8648d1125872ab30bcfaf40181e986c166209360a32072f14265c0",
+    "validation/stages/path_diagnostics.py": "1da02f7eee2b9da74476f7d8e5c46bc6cf23b760d87be519e8c1c82fb9b92ba2",
+    "validation/stages/prepare_review_tasks.py": "9e57a0d1f6d2b61c2b1d2879c81304d5cbb0afed00b3b5b599f196a028bbd85e",
+    "validation/stages/reference_leaks.py": "1b9817d0c070a19b0bdc5365c5614ac335da6156cdcee61e80332e16786b7cc1",
+    "validation/stages/run_reused_containers.py": "cb5f2ee51354d1a83ad2951cfe67d596331cd87dfb6399eba4ec4e1255b3eac7",
+    "validation/stages/runner.py": "825ce76be519c63592f9a5305203512256d68adf849d2efb25399dd71b32c8d8",
+    "validation/stages/service_entrypoint.py": "e328bb86ee3f49a2f3a5d4043d04cb4bc22403417cc07681c18d22d78220bd2f",
+    "validation/stages/task_setup.py": "8de3f30f61f0167785b66f68a642854ae68ee7667450d2c2a99cdfdf0509338c",
+    "validation/stages/verifier_setup.py": "051a80e170b933230ad4fd0e1459f44bd9901e2b399e388a9959c2fc73358517",
+    "validation/upstream/__init__.py": "31f6f89e99cd288755413fc80aeb983b9dcde796a2624d80b6cd2c07315a86c9",
+    "validation/upstream/__main__.py": "fb76f9492685add6774fdf9faf6ee0cffcd78c5860727b5c671351429b484194",
+    "validation/upstream/terminal_bench/scripts/checks/check_ai_detection.py": "3cdcdc9229f9be6d2d7025fb179295be2039a76a9cac2b89228a85197f15d9cb"
+  },
+  "arguments": {
+    "tasks": "/hnvme/workspace/y500bb12-optiagent/fineenvs-v2-20261026/publication/fineenvs_v2/archive.parquet",
+    "pack_image_cache": false,
+    "fix_instruction_paths": true,
+    "fix_pip_pins": false,
+    "path_resolution_report": [],
+    "fix_repository_leaks": false,
+    "fix_instruction_suffix": false,
+    "dataset_source": "FWeindel/validated-tasks/fineenvs_v2",
+    "dataset_revision": "6dcbab71a3926fe61a5d630a8e7f64edb16f1a54",
+    "min_tasks": 999,
+    "out": "/hnvme/workspace/y500bb12-optiagent/fineenvs-qwen38-thinking-20261026/results",
+    "backend": "apptainer",
+    "environment_kwargs": {},
+    "agent": "terminus-2",
+    "review_evidence": "image",
+    "review_max_budget_usd": 5,
+    "review_reasoning_effort": "medium",
+    "review_agent": null,
+    "model": null,
+    "review_model": null,
+    "review_pricing": null,
+    "review_rubric": "terminal-bench",
+    "review_format": null,
+    "review_criteria": null,
+    "review_local": false,
+    "analysis_agent": null,
+    "analysis_model": null,
+    "agent_kwargs": {
+      "temperature": 1.0,
+      "max_tokens": 8192,
+      "llm_call_kwargs": {
+        "max_tokens": 8192
+      },
+      "extra_body": {
+        "top_p": 0.95,
+        "top_k": 20,
+        "min_p": 0,
+        "presence_penalty": 0,
+        "repetition_penalty": 1.0,
+        "chat_template_kwargs": {
+          "enable_thinking": true
+        }
+      }
+    },
+    "attempts": 2,
+    "review_runs": null,
+    "review_retries": 2,
+    "review_resume": true,
+    "review_seed": 42,
+    "outcome_retries": 3,
+    "expected_outcomes": null,
+    "review_setup": null,
+    "preparation_runs": 1,
+    "preparation_mean_target_seconds": null,
+    "preparation_median_target_seconds": null,
+    "preparation_max_seconds": null,
+    "concurrency": 4,
+    "static_concurrency": null,
+    "reuse_stage3": null,
+    "static_resume": null,
+    "static_resume_accept_previous_path_check": false,
+    "reuse_validation_containers": false,
+    "container_start_concurrency": 4,
+    "image_build_memory_mb": 8192,
+    "image_build_cpus": 4,
+    "image_build_concurrency": 2,
+    "image_build_timeout_sec": 3600,
+    "prebuild_only": false,
+    "dependency_archives": null,
+    "dependency_layout": null,
+    "container_start_interval": 0,
+    "limit": null,
+    "task_id_range": null,
+    "static_profile": "training",
+    "resolve_path_root": [],
+    "exclude": [],
+    "submit": "auto",
+    "time": "08:00:00",
+    "partition": "h200",
+    "cpus": 32,
+    "gpus": null,
+    "memory": null,
+    "trial_cpus": null,
+    "trial_memory_mb": null,
+    "serve_model": "qwen38-27b",
+    "serve_weights": "/hnvme/workspace/y500bb12-optiagent/models/Qwen3.8-27B",
+    "serve_context": 32768,
+    "serve_replicas": 1,
+    "serve_cpus": 4,
+    "serve_memory_mb": 32768,
+    "max_num_seqs": 8,
+    "api_base": null,
+    "network_mode": "isolated",
+    "reward_key": "reward",
+    "trials": null,
+    "max_iterations": 10,
+    "timeout_minutes": 30,
+    "publish_repo": null,
+    "publish_require_complete": false,
+    "publish_folder": [],
+    "publish_readme": false,
+    "publish_readme_force": false,
+    "publish_readme_model": "claude-fable-5-1",
+    "publish_readme_seed": 0,
+    "publish_readme_evidence": [],
+    "publish_analysis": false,
+    "publish_patch_manifest": [],
+    "publish_conversion_archive": [],
+    "force_build": false,
+    "dry_run": false
+  },
+  "task_manifest": {
+    "path": "contract.tasks.json",
+    "count": 999,
+    "sha256": "9b9b39261e8dfb616ccfc5d2a67ae00f048c6ce27c1432bc76a7d591ecce1459"
+  },
+  "sha256": "9a7c288630f67227220fd23a62bc0978f4db02623b18c6e03d84d10fff09d937"
+}

fineenvs_v3/metadata/evaluation-settings-audit.json

@@ -0,0 +1,125 @@
+{
+  "scope": "All 1,998 saved trial configurations (result.json config.agent.kwargs; api_base removed).",
+  "configs": [
+    {
+      "attempts": 1998,
+      "kwargs": {
+        "agent": "terminus-2",
+        "extra_body": {
+          "chat_template_kwargs": {
+            "enable_thinking": true
+          },
+          "min_p": 0,
+          "presence_penalty": 0,
+          "repetition_penalty": 1.0,
+          "top_k": 20,
+          "top_p": 0.95
+        },
+        "llm_call_kwargs": {
+          "max_tokens": 8192
+        },
+        "max_tokens": 8192,
+        "model_info": {
+          "input_cost_per_token": 0,
+          "max_input_tokens": 32768,
+          "max_output_tokens": 8192,
+          "output_cost_per_token": 0
+        },
+        "model_name": "openai/qwen38-27b",
+        "temperature": 1.0
+      }
+    }
+  ],
+  "timeouts": [
+    {
+      "attempts": 1972,
+      "limits": {
+        "agent": 600.0,
+        "agent_setup": 360.0,
+        "environment_build": 600.0,
+        "steps": [],
+        "trial_attempt": null,
+        "trial_cleanup_grace": 30.0,
+        "verifier": 120.0
+      }
+    },
+    {
+      "attempts": 26,
+      "limits": {
+        "agent": 900.0,
+        "agent_setup": 360.0,
+        "environment_build": 600.0,
+        "steps": [],
+        "trial_attempt": null,
+        "trial_cleanup_grace": 30.0,
+        "verifier": 120.0
+      }
+    }
+  ],
+  "evaluation_settings": [
+    {
+      "agent": "terminus-2",
+      "agent_timeout": "task-specific; see saved task.toml and resolved trial timeouts",
+      "attempts": 2,
+      "context_tokens": 32768,
+      "contract_sha256": "9a7c288630f67227220fd23a62bc0978f4db02623b18c6e03d84d10fff09d937",
+      "hf_repo": "Qwen/Qwen3.8-27B",
+      "intended_output_tokens_per_call": 8192,
+      "model": "qwen38-27b",
+      "output_limit_status": "present in saved request kwargs; HTTP payload not audited",
+      "request_output_tokens": 8192,
+      "resolved_timeouts": {
+        "agent": 600.0,
+        "agent_setup": 360.0,
+        "environment_build": 600.0,
+        "steps": [],
+        "trial_attempt": null,
+        "trial_cleanup_grace": 30.0,
+        "verifier": 120.0
+      },
+      "sampling": {
+        "min_p": 0,
+        "presence_penalty": 0,
+        "repetition_penalty": 1.0,
+        "temperature": 1.0,
+        "top_k": 20,
+        "top_p": 0.95
+      },
+      "thinking": true,
+      "weights_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+      "attempts_covered": 1972
+    },
+    {
+      "agent": "terminus-2",
+      "agent_timeout": "task-specific; see saved task.toml and resolved trial timeouts",
+      "attempts": 2,
+      "context_tokens": 32768,
+      "contract_sha256": "9a7c288630f67227220fd23a62bc0978f4db02623b18c6e03d84d10fff09d937",
+      "hf_repo": "Qwen/Qwen3.8-27B",
+      "intended_output_tokens_per_call": 8192,
+      "model": "qwen38-27b",
+      "output_limit_status": "present in saved request kwargs; HTTP payload not audited",
+      "request_output_tokens": 8192,
+      "resolved_timeouts": {
+        "agent": 900.0,
+        "agent_setup": 360.0,
+        "environment_build": 600.0,
+        "steps": [],
+        "trial_attempt": null,
+        "trial_cleanup_grace": 30.0,
+        "verifier": 120.0
+      },
+      "sampling": {
+        "min_p": 0,
+        "presence_penalty": 0,
+        "repetition_penalty": 1.0,
+        "temperature": 1.0,
+        "top_k": 20,
+        "top_p": 0.95
+      },
+      "thinking": true,
+      "weights_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+      "attempts_covered": 26
+    }
+  ]
+}

fineenvs_v3/metadata/release-verification.json

@@ -0,0 +1,38 @@
+{
+  "files": {
+    "tasks.parquet": {
+      "rows": 515,
+      "sha256": "9f30c3d403ff8a3dd8e4e368897386239a6c4ab334e9484fd5ded204b87b0c46",
+      "payload_hashes_verified": true,
+      "payload_columns_identical_to_evaluated_source": true
+    },
+    "archive.parquet": {
+      "rows": 484,
+      "sha256": "a7a1473e1851abc0b3648f266ee009ada1725219b97bedb919cc894a3fe46734",
+      "payload_hashes_verified": true,
+      "payload_columns_identical_to_evaluated_source": true
+    }
+  },
+  "evaluated_source": {
+    "file": "fineenvs_v2/archive.parquet",
+    "sha256": "fbfee6fd39341d24057ff4b355de7a0c4b91923faa17fb4ea8bdded42b6c58b8",
+    "revision": "d6372d820796588bfa45f70010d1e0bf470b70eb"
+  },
+  "patch_py_reproduces_rows_and_metadata": true,
+  "no_v2_task_overlap": true,
+  "trace_attempts": 1998,
+  "saved_trajectories": 1998,
+  "rewards_match_stage6": true,
+  "reward_origins": {
+    "verifier": 1992,
+    "terminal_error": 6
+  },
+  "exceptions": {
+    "none": 1924,
+    "AgentTimeoutError": 67,
+    "RepeatedOutputTruncationError": 1,
+    "TmuxBatchProtocolError": 6
+  },
+  "credential_pattern_hits": 0,
+  "network": "isolation requested; all 40 job allocations fell back to the host network (max_net_namespaces exceeded)"
+}

fineenvs_v3/metadata/results.json

@@ -0,0 +1,66 @@
+{
+  "model": "Qwen/Qwen3.8-27B",
+  "thinking": true,
+  "evaluated": 999,
+  "kept": 515,
+  "still_unsolved": 484,
+  "v2_kept": 4001,
+  "combined_kept": 4516,
+  "attempts": 1998,
+  "successful_attempts": 897,
+  "saved_trajectories": 1998,
+  "null_verifier_rewards": 6,
+  "distribution": {
+    "0/2": 484,
+    "1/2": 133,
+    "2/2": 382
+  },
+  "pass@1": 0.44894894894894893,
+  "pass@2": 0.5155155155155156,
+  "reward_policy": "terminal-errors-zero-v2",
+  "layout": "incremental",
+  "attempts_without_verifier_reward": [
+    {
+      "task_id": "0045_605_45605559_qa_4",
+      "attempt_id": "6366b81d-d598-4015-865b-ed2f416c6661",
+      "reward": 0,
+      "reward_origin": "terminal_error",
+      "exception": "TmuxBatchProtocolError"
+    },
+    {
+      "task_id": "0045_605_45605559_qa_4",
+      "attempt_id": "179b8a70-04e7-4222-b709-63446645fe4a",
+      "reward": 0,
+      "reward_origin": "terminal_error",
+      "exception": "TmuxBatchProtocolError"
+    },
+    {
+      "task_id": "0135_735_135735954_qa_4",
+      "attempt_id": "3b02780d-e194-4ea5-a5a2-736eca2b7758",
+      "reward": 0,
+      "reward_origin": "terminal_error",
+      "exception": "TmuxBatchProtocolError"
+    },
+    {
+      "task_id": "0120_574_120574465_qa_1",
+      "attempt_id": "6bb93a6b-dcfb-49e0-ad2f-b99c8a81ce13",
+      "reward": 0,
+      "reward_origin": "terminal_error",
+      "exception": "TmuxBatchProtocolError"
+    },
+    {
+      "task_id": "0051_855_51855195_qa_2",
+      "attempt_id": "f5ee21b8-49ff-4d97-8397-9554375c0903",
+      "reward": 0,
+      "reward_origin": "terminal_error",
+      "exception": "TmuxBatchProtocolError"
+    },
+    {
+      "task_id": "0055_859_55859289_qa_2",
+      "attempt_id": "e9fb6a10-c02c-409d-8485-8d8818d9a370",
+      "reward": 0,
+      "reward_origin": "terminal_error",
+      "exception": "TmuxBatchProtocolError"
+    }
+  ]
+}

fineenvs_v3/metadata/stage-7-trace-metrics.json

@@ -0,0 +1,783 @@
+{
+  "note": "Recomputed from all 40 shard stage-7 reports with validation.checks.trace_metrics.aggregate; stage-7-merged.json trace_metrics holds shard 0 only. Model-call latency is omitted because per-call times are not kept in saved rows; per-shard values remain in each shard report. No throughput (shards overlap).",
+  "shards": 40,
+  "tasks_count": 999,
+  "trajectories_per_task": {
+    "2": 999
+  },
+  "all_tasks": {
+    "trajectories": 1998,
+    "graded": 1998,
+    "solved": 897,
+    "turns": {
+      "n": 1998,
+      "mean": 5.594094094094094,
+      "median": 5,
+      "p90": 9,
+      "p95": 11,
+      "max": 20
+    },
+    "input_tokens": {
+      "n": 1998,
+      "mean": 24375.874374374373,
+      "median": 13384,
+      "p90": 57469,
+      "p95": 87635,
+      "max": 254708
+    },
+    "input_tokens_total": 48702997,
+    "output_tokens": {
+      "n": 1998,
+      "mean": 4956.957457457457,
+      "median": 2774,
+      "p90": 12053,
+      "p95": 17458,
+      "max": 36812
+    },
+    "output_tokens_total": 9904001,
+    "peak_context_tokens": {
+      "n": 1998,
+      "mean": 6639.290290290291,
+      "median": 4604,
+      "p90": 13212,
+      "p95": 17950,
+      "max": 46457
+    },
+    "peak_context_fraction": {
+      "n": 1998,
+      "mean": 0.20261506012848787,
+      "median": 0.1405029296875,
+      "p90": 0.4031982421875,
+      "p95": 0.54779052734375,
+      "max": 1.417755126953125
+    },
+    "trajectories_above_90_percent_context": 13,
+    "latency_seconds": {
+      "model_call": null,
+      "tool_call": null,
+      "agent": {
+        "n": 1998,
+        "mean": 123.11313489439439,
+        "median": 61.362441,
+        "p90": 316.901775,
+        "p95": 483.252097,
+        "max": 603.230676
+      },
+      "verifier": {
+        "n": 1992,
+        "mean": 8.583377089859438,
+        "median": 8.445156,
+        "p90": 9.493342,
+        "p95": 10.229738,
+        "max": 23.125993
+      },
+      "trial": {
+        "n": 1998,
+        "mean": 143.24902637437438,
+        "median": 81.229523,
+        "p90": 337.991541,
+        "p95": 503.594045,
+        "max": 764.450748
+      }
+    },
+    "tools": {
+      "bash_command": {
+        "calls": 13590,
+        "calls_per_trajectory": 6.801801801801802,
+        "failed": null,
+        "failure_rate": null
+      },
+      "mark_task_complete": {
+        "calls": 3995,
+        "calls_per_trajectory": 1.9994994994994995,
+        "failed": null,
+        "failure_rate": null
+      }
+    },
+    "tool_steps": 11022,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 789,
+    "tool_steps_with_error_text_rate": 0.07158410451823626,
+    "terminations": {
+      "environment_error": 6,
+      "repeated_output_truncation": 1,
+      "task_complete": 1924,
+      "task_timeout": 67
+    },
+    "command_distribution": {
+      "[assignment]": 0.00031638930937078456,
+      "[unknown]": 0.0018553259204475904,
+      "apt": 1.6145177435500014e-05,
+      "apt-cache": 1.668335001668335e-05,
+      "apt-get": 0.00014091602292706624,
+      "awk": 0.0006099950206940828,
+      "cat": 0.16885301963055835,
+      "cd": 0.0737891846309199,
+      "cp": 5.98000598000598e-05,
+      "cut": 0.0005676251419285476,
+      "df": 3.0860900426117815e-05,
+      "diff": 3.707411114818522e-05,
+      "echo": 0.14505262385744908,
+      "export": 3.33667000333667e-05,
+      "file": 0.0014380270123773597,
+      "find": 0.007022134159026405,
+      "free": 0.00016538070388901857,
+      "grep": 0.008289016329971937,
+      "head": 0.09729647621978928,
+      "hexdump": 5.005005005005005e-05,
+      "jobs": 1.82000182000182e-05,
+      "kill": 9.443405669820763e-06,
+      "ldconfig": 2.86000286000286e-05,
+      "ln": 1.668335001668335e-05,
+      "ls": 0.07504878078790124,
+      "md5sum": 1.7875017875017873e-05,
+      "mkdir": 0.08327001484715718,
+      "mv": 0.00022720720788353642,
+      "nproc": 8.780910903552414e-05,
+      "od": 0.000552732932907225,
+      "pip": 0.0026554174009669777,
+      "printf": 0.0012706301079194095,
+      "ps": 2.275002275002275e-05,
+      "pwd": 0.0008553844306650322,
+      "python": 0.28992040946265074,
+      "python-m:pip": 0.00013485707930152374,
+      "q": 2.275002275002275e-05,
+      "rm": 9.753110598521226e-05,
+      "sed": 0.0014005408739936256,
+      "sha256sum": 1.0214295928581642e-05,
+      "sleep": 7.563583553366441e-05,
+      "sort": 0.000415621090434352,
+      "sqlite3": 0.0007315636400603722,
+      "strings": 2.7679194345861013e-05,
+      "tail": 0.0042530803760999405,
+      "tee": 0.00011560473516995256,
+      "timeout": 0.00032037847485169666,
+      "tr": 0.00027212846024711533,
+      "true": 0.00042095319873097645,
+      "uniq": 0.00011836723793245531,
+      "wc": 0.030980963419687303,
+      "which": 0.0003818396332624189,
+      "xargs": 9.175842509175844e-05,
+      "xxd": 0.00048650046213399037
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 32,
+    "trajectories_with_oom_recovery": 26,
+    "malformed_tool_calls": 81,
+    "trajectories_with_malformed_tool_calls": 76,
+    "errors": {
+      "model_server": {
+        "trajectories": 0,
+        "types": {}
+      },
+      "environment": {
+        "trajectories": 6,
+        "types": {
+          "TmuxBatchProtocolError": 6
+        }
+      },
+      "verifier": {
+        "trajectories": 0,
+        "types": {}
+      },
+      "other": {
+        "trajectories": 0,
+        "types": {}
+      }
+    }
+  },
+  "families": {
+    "all": {
+      "trajectories": 1998,
+      "graded": 1998,
+      "solved": 897,
+      "turns": {
+        "n": 1998,
+        "mean": 5.594094094094094,
+        "median": 5,
+        "p90": 9,
+        "p95": 11,
+        "max": 20
+      },
+      "input_tokens": {
+        "n": 1998,
+        "mean": 24375.874374374373,
+        "median": 13384,
+        "p90": 57469,
+        "p95": 87635,
+        "max": 254708
+      },
+      "input_tokens_total": 48702997,
+      "output_tokens": {
+        "n": 1998,
+        "mean": 4956.957457457457,
+        "median": 2774,
+        "p90": 12053,
+        "p95": 17458,
+        "max": 36812
+      },
+      "output_tokens_total": 9904001,
+      "peak_context_tokens": {
+        "n": 1998,
+        "mean": 6639.290290290291,
+        "median": 4604,
+        "p90": 13212,
+        "p95": 17950,
+        "max": 46457
+      },
+      "peak_context_fraction": {
+        "n": 1998,
+        "mean": 0.20261506012848787,
+        "median": 0.1405029296875,
+        "p90": 0.4031982421875,
+        "p95": 0.54779052734375,
+        "max": 1.417755126953125
+      },
+      "trajectories_above_90_percent_context": 13,
+      "latency_seconds": {
+        "model_call": null,
+        "tool_call": null,
+        "agent": {
+          "n": 1998,
+          "mean": 123.11313489439439,
+          "median": 61.362441,
+          "p90": 316.901775,
+          "p95": 483.252097,
+          "max": 603.230676
+        },
+        "verifier": {
+          "n": 1992,
+          "mean": 8.583377089859438,
+          "median": 8.445156,
+          "p90": 9.493342,
+          "p95": 10.229738,
+          "max": 23.125993
+        },
+        "trial": {
+          "n": 1998,
+          "mean": 143.24902637437438,
+          "median": 81.229523,
+          "p90": 337.991541,
+          "p95": 503.594045,
+          "max": 764.450748
+        }
+      },
+      "tools": {
+        "bash_command": {
+          "calls": 13590,
+          "calls_per_trajectory": 6.801801801801802,
+          "failed": null,
+          "failure_rate": null
+        },
+        "mark_task_complete": {
+          "calls": 3995,
+          "calls_per_trajectory": 1.9994994994994995,
+          "failed": null,
+          "failure_rate": null
+        }
+      },
+      "tool_steps": 11022,
+      "tool_steps_without_output": 0,
+      "tool_steps_with_error_text": 789,
+      "tool_steps_with_error_text_rate": 0.07158410451823626,
+      "terminations": {
+        "environment_error": 6,
+        "repeated_output_truncation": 1,
+        "task_complete": 1924,
+        "task_timeout": 67
+      },
+      "command_distribution": {
+        "[assignment]": 0.00031638930937078456,
+        "[unknown]": 0.0018553259204475904,
+        "apt": 1.6145177435500014e-05,
+        "apt-cache": 1.668335001668335e-05,
+        "apt-get": 0.00014091602292706624,
+        "awk": 0.0006099950206940828,
+        "cat": 0.16885301963055835,
+        "cd": 0.0737891846309199,
+        "cp": 5.98000598000598e-05,
+        "cut": 0.0005676251419285476,
+        "df": 3.0860900426117815e-05,
+        "diff": 3.707411114818522e-05,
+        "echo": 0.14505262385744908,
+        "export": 3.33667000333667e-05,
+        "file": 0.0014380270123773597,
+        "find": 0.007022134159026405,
+        "free": 0.00016538070388901857,
+        "grep": 0.008289016329971937,
+        "head": 0.09729647621978928,
+        "hexdump": 5.005005005005005e-05,
+        "jobs": 1.82000182000182e-05,
+        "kill": 9.443405669820763e-06,
+        "ldconfig": 2.86000286000286e-05,
+        "ln": 1.668335001668335e-05,
+        "ls": 0.07504878078790124,
+        "md5sum": 1.7875017875017873e-05,
+        "mkdir": 0.08327001484715718,
+        "mv": 0.00022720720788353642,
+        "nproc": 8.780910903552414e-05,
+        "od": 0.000552732932907225,
+        "pip": 0.0026554174009669777,
+        "printf": 0.0012706301079194095,
+        "ps": 2.275002275002275e-05,
+        "pwd": 0.0008553844306650322,
+        "python": 0.28992040946265074,
+        "python-m:pip": 0.00013485707930152374,
+        "q": 2.275002275002275e-05,
+        "rm": 9.753110598521226e-05,
+        "sed": 0.0014005408739936256,
+        "sha256sum": 1.0214295928581642e-05,
+        "sleep": 7.563583553366441e-05,
+        "sort": 0.000415621090434352,
+        "sqlite3": 0.0007315636400603722,
+        "strings": 2.7679194345861013e-05,
+        "tail": 0.0042530803760999405,
+        "tee": 0.00011560473516995256,
+        "timeout": 0.00032037847485169666,
+        "tr": 0.00027212846024711533,
+        "true": 0.00042095319873097645,
+        "uniq": 0.00011836723793245531,
+        "wc": 0.030980963419687303,
+        "which": 0.0003818396332624189,
+        "xargs": 9.175842509175844e-05,
+        "xxd": 0.00048650046213399037
+      },
+      "output_cap_events": 0,
+      "oom_recoveries": 32,
+      "trajectories_with_oom_recovery": 26,
+      "malformed_tool_calls": 81,
+      "trajectories_with_malformed_tool_calls": 76,
+      "errors": {
+        "model_server": {
+          "trajectories": 0,
+          "types": {}
+        },
+        "environment": {
+          "trajectories": 6,
+          "types": {
+            "TmuxBatchProtocolError": 6
+          }
+        },
+        "verifier": {
+          "trajectories": 0,
+          "types": {}
+        },
+        "other": {
+          "trajectories": 0,
+          "types": {}
+        }
+      }
+    }
+  },
+  "model_server_requests": {
+    "requests": 11229,
+    "failed": 4,
+    "failure_rate": 0.00035622050048980317,
+    "by_status": {
+      "200": 11225,
+      "400": 4
+    }
+  },
+  "inference_resources_by_shard": {
+    "0000": {
+      "0": {
+        "peak_memory_mib": 128755.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 188,
+        "mean_memory_mib": 128360.59574468085,
+        "mean_utilization_percent": 79.24468085106383
+      }
+    },
+    "0001": {
+      "0": {
+        "peak_memory_mib": 128823.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 230,
+        "mean_memory_mib": 128481.82608695653,
+        "mean_utilization_percent": 88.69565217391305
+      }
+    },
+    "0002": {
+      "0": {
+        "peak_memory_mib": 129865.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 178,
+        "mean_memory_mib": 129441.01123595505,
+        "mean_utilization_percent": 89.82022471910112
+      }
+    },
+    "0003": {
+      "0": {
+        "peak_memory_mib": 129433.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 191,
+        "mean_memory_mib": 128637.69109947643,
+        "mean_utilization_percent": 90.39267015706807
+      }
+    },
+    "0004": {
+      "0": {
+        "peak_memory_mib": 129439.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 292,
+        "mean_memory_mib": 128703.58219178082,
+        "mean_utilization_percent": 92.0068493150685
+      }
+    },
+    "0005": {
+      "0": {
+        "peak_memory_mib": 128653.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 263,
+        "mean_memory_mib": 128463.25855513308,
+        "mean_utilization_percent": 86.95817490494296
+      }
+    },
+    "0006": {
+      "0": {
+        "peak_memory_mib": 129535.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 280,
+        "mean_memory_mib": 129443.35,
+        "mean_utilization_percent": 92.64285714285714
+      }
+    },
+    "0007": {
+      "0": {
+        "peak_memory_mib": 128859.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 175,
+        "mean_memory_mib": 128392.14285714286,
+        "mean_utilization_percent": 78.41714285714286
+      }
+    },
+    "0008": {
+      "0": {
+        "peak_memory_mib": 129349.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 191,
+        "mean_memory_mib": 129088.19371727749,
+        "mean_utilization_percent": 87.43455497382199
+      }
+    },
+    "0009": {
+      "0": {
+        "peak_memory_mib": 128775.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 223,
+        "mean_memory_mib": 128697.10762331838,
+        "mean_utilization_percent": 73.02690582959642
+      }
+    },
+    "0010": {
+      "0": {
+        "peak_memory_mib": 128555.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 147,
+        "mean_memory_mib": 128360.6462585034,
+        "mean_utilization_percent": 86.37414965986395
+      }
+    },
+    "0011": {
+      "0": {
+        "peak_memory_mib": 128551.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 225,
+        "mean_memory_mib": 128427.25777777778,
+        "mean_utilization_percent": 86.83111111111111
+      }
+    },
+    "0012": {
+      "0": {
+        "peak_memory_mib": 129635.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 161,
+        "mean_memory_mib": 128579.54658385093,
+        "mean_utilization_percent": 81.98757763975155
+      }
+    },
+    "0013": {
+      "0": {
+        "peak_memory_mib": 128533.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 183,
+        "mean_memory_mib": 128385.05464480874,
+        "mean_utilization_percent": 85.24590163934427
+      }
+    },
+    "0014": {
+      "0": {
+        "peak_memory_mib": 129383.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 304,
+        "mean_memory_mib": 128690.08552631579,
+        "mean_utilization_percent": 89.4342105263158
+      }
+    },
+    "0015": {
+      "0": {
+        "peak_memory_mib": 128719.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 133,
+        "mean_memory_mib": 128130.02255639098,
+        "mean_utilization_percent": 79.67669172932331
+      }
+    },
+    "0016": {
+      "0": {
+        "peak_memory_mib": 128873.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 220,
+        "mean_memory_mib": 128782.27272727272,
+        "mean_utilization_percent": 81.31363636363636
+      }
+    },
+    "0017": {
+      "0": {
+        "peak_memory_mib": 129633.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 176,
+        "mean_memory_mib": 129030.26136363637,
+        "mean_utilization_percent": 89.10227272727273
+      }
+    },
+    "0018": {
+      "0": {
+        "peak_memory_mib": 129077.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 229,
+        "mean_memory_mib": 128805.98689956332,
+        "mean_utilization_percent": 88.17903930131004
+      }
+    },
+    "0019": {
+      "0": {
+        "peak_memory_mib": 129351.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 212,
+        "mean_memory_mib": 128891.85849056604,
+        "mean_utilization_percent": 86.79245283018868
+      }
+    },
+    "0020": {
+      "0": {
+        "peak_memory_mib": 128567.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 190,
+        "mean_memory_mib": 128421.27368421052,
+        "mean_utilization_percent": 87.13684210526316
+      }
+    },
+    "0021": {
+      "0": {
+        "peak_memory_mib": 129353.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 176,
+        "mean_memory_mib": 129162.73863636363,
+        "mean_utilization_percent": 92.39204545454545
+      }
+    },
+    "0022": {
+      "0": {
+        "peak_memory_mib": 129173.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 208,
+        "mean_memory_mib": 128819.27884615384,
+        "mean_utilization_percent": 90.86538461538461
+      }
+    },
+    "0023": {
+      "0": {
+        "peak_memory_mib": 129869.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 191,
+        "mean_memory_mib": 128931.30366492146,
+        "mean_utilization_percent": 87.42408376963351
+      }
+    },
+    "0024": {
+      "0": {
+        "peak_memory_mib": 128891.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 238,
+        "mean_memory_mib": 128807.9243697479,
+        "mean_utilization_percent": 86.1218487394958
+      }
+    },
+    "0025": {
+      "0": {
+        "peak_memory_mib": 128945.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 233,
+        "mean_memory_mib": 128390.16738197424,
+        "mean_utilization_percent": 80.02145922746782
+      }
+    },
+    "0026": {
+      "0": {
+        "peak_memory_mib": 128169.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 179,
+        "mean_memory_mib": 128060.22905027933,
+        "mean_utilization_percent": 85.92178770949721
+      }
+    },
+    "0027": {
+      "0": {
+        "peak_memory_mib": 128701.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 199,
+        "mean_memory_mib": 128530.26633165829,
+        "mean_utilization_percent": 89.24120603015075
+      }
+    },
+    "0028": {
+      "0": {
+        "peak_memory_mib": 129177.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 233,
+        "mean_memory_mib": 129081.90128755364,
+        "mean_utilization_percent": 79.79828326180258
+      }
+    },
+    "0029": {
+      "0": {
+        "peak_memory_mib": 128827.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 168,
+        "mean_memory_mib": 128435.88095238095,
+        "mean_utilization_percent": 87.76190476190476
+      }
+    },
+    "0030": {
+      "0": {
+        "peak_memory_mib": 128293.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 159,
+        "mean_memory_mib": 128155.83018867925,
+        "mean_utilization_percent": 89.74842767295597
+      }
+    },
+    "0031": {
+      "0": {
+        "peak_memory_mib": 129087.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 161,
+        "mean_memory_mib": 128787.0745341615,
+        "mean_utilization_percent": 80.8695652173913
+      }
+    },
+    "0032": {
+      "0": {
+        "peak_memory_mib": 129431.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 279,
+        "mean_memory_mib": 128818.39784946236,
+        "mean_utilization_percent": 84.72043010752688
+      }
+    },
+    "0033": {
+      "0": {
+        "peak_memory_mib": 129167.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 135,
+        "mean_memory_mib": 128656.18518518518,
+        "mean_utilization_percent": 88.36296296296297
+      }
+    },
+    "0034": {
+      "0": {
+        "peak_memory_mib": 129191.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 216,
+        "mean_memory_mib": 128427.0925925926,
+        "mean_utilization_percent": 90.67129629629629
+      }
+    },
+    "0035": {
+      "0": {
+        "peak_memory_mib": 128495.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 261,
+        "mean_memory_mib": 128386.38697318008,
+        "mean_utilization_percent": 89.57471264367815
+      }
+    },
+    "0036": {
+      "0": {
+        "peak_memory_mib": 128827.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 219,
+        "mean_memory_mib": 128465.61187214612,
+        "mean_utilization_percent": 88.58447488584476
+      }
+    },
+    "0037": {
+      "0": {
+        "peak_memory_mib": 128569.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 233,
+        "mean_memory_mib": 128503.48927038627,
+        "mean_utilization_percent": 86.61802575107296
+      }
+    },
+    "0038": {
+      "0": {
+        "peak_memory_mib": 129111.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 210,
+        "mean_memory_mib": 128988.6380952381,
+        "mean_utilization_percent": 82.46190476190476
+      }
+    },
+    "0039": {
+      "0": {
+        "peak_memory_mib": 129279.0,
+        "memory_total_mib": 143771.0,
+        "peak_utilization_percent": 100.0,
+        "samples": 208,
+        "mean_memory_mib": 128800.32692307692,
+        "mean_utilization_percent": 84.61538461538461
+      }
+    }
+  }
+}

fineenvs_v3/patch.py

@@ -0,0 +1,45 @@
+"""Reproduce FineEnvs v3 membership: newly solved tasks and the remaining-unsolved archive.
+python patch.py --out reproduced [--source /path/to/pinned/fineenvs_v2/archive.parquet]
+Requires pyarrow and huggingface_hub; metadata/selection.json beside this script. Task payloads
+are the evaluated fineenvs_v2 archive rows; only archive status and success-count columns change.
+Does not rerun evaluation.
+"""
+from pathlib import Path
+import argparse,json,hashlib
+import pyarrow as pa,pyarrow.parquet as pq
+
+REASON='No observed success in four Qwen/Qwen3-30B-A3B-Instruct-2507 attempts (v2) or two Qwen/Qwen3.8-27B thinking attempts (v3); terminal errors counted as unsolved.'
+
+def main():
+ p=argparse.ArgumentParser();p.add_argument('--out',type=Path,required=True);p.add_argument('--source',type=Path);a=p.parse_args()
+ s=json.loads((Path(__file__).parent/'metadata/selection.json').read_text());tasks=s['tasks']
+ if a.source is None:
+  from huggingface_hub import hf_hub_download
+  a.source=Path(hf_hub_download('FWeindel/validated-tasks',f"{s['source_folder']}/{s['source_file']}",repo_type='dataset',revision=s['source_revision']))
+ assert hashlib.file_digest(a.source.open('rb'),'sha256').hexdigest()==s['source_sha256']
+ pf=pq.ParquetFile(a.source);schema=pf.schema_arrow
+ for c,typ in [('qwen38_successes',pa.int64()),('qwen38_attempts',pa.int64()),('kept_from_v2',pa.bool_())]:schema=schema.append(pa.field(c,typ))
+ kept={t for t,v in tasks.items() if v['kept']};a.out.mkdir(parents=True,exist_ok=True);writers={}
+ for name,chosen in [('tasks.parquet',kept),('archive.parquet',set(tasks)-kept)]:
+  meta=dict(schema.metadata or {});im=json.loads(meta[b'ot.images.v1'])
+  im['tasks']={t:v for t,v in im['tasks'].items() if t in chosen};used={e['bundle'] for v in im['tasks'].values() for e in v['environments']};im['bundles']={k:v for k,v in im['bundles'].items() if k in used}
+  if 'coverage' in im:im['coverage']={**im['coverage'],'cached_tasks':len(im['tasks']),'total_tasks':len(chosen)}
+  if 'uncached_task_ids' in im:im['uncached_task_ids']=sorted(chosen-set(im['tasks']))
+  meta[b'ot.images.v1']=json.dumps(im).encode()
+  meta[b'ot.release.v3']=json.dumps({'layout':'incremental','kept':len(kept),'rows':len(chosen),'baseline_folder':s['baseline_folder'],'baseline_revision':s['source_revision'],'source':f"{s['source_folder']}/{s['source_file']}",'source_sha256':s['source_sha256']}).encode()
+  writers[name]=pq.ParquetWriter(a.out/name,schema.with_metadata(meta),compression='zstd')
+ seen=set()
+ try:
+  for batch in pf.iter_batches(batch_size=4):
+   groups={n:[] for n in writers}
+   for row in batch.to_pylist():
+    v=tasks[row['path']];assert v['content_sha256']==row['content_sha256'];seen.add(row['path'])
+    if v['kept']:row.update(archive_stage=None,archive_reason=None,archive_labels=[]);n='tasks.parquet'
+    else:row.update(archive_stage=6,archive_reason=REASON,archive_labels=['teacher:no-observed-success']);n='archive.parquet'
+    row.update(qwen38_successes=v['qwen38_successes'],qwen38_attempts=v['qwen38_attempts'],kept_from_v2=False);groups[n].append(row)
+   for n,rows in groups.items():
+    if rows:writers[n].write_table(pa.Table.from_pylist(rows,schema=writers[n].schema))
+ finally:
+  for w in writers.values():w.close()
+ assert seen==set(tasks)
+if __name__=='__main__':main()

fineenvs_v3/tasks.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f30c3d403ff8a3dd8e4e368897386239a6c4ab334e9484fd5ded204b87b0c46
+size 4660009682

fineenvs_v3/trace_stats/README.md

@@ -0,0 +1,57 @@
+# FineEnvs v3 trace statistics
+
+## Coverage
+
+Of 1,998 recorded attempts, 1,998 have saved conversations and 0 do not. 6 attempts have no raw verifier reward. Missing measurements remain missing.
+
+Agent timeouts and verifier outcomes are separate: an attempt can time out and still receive a successful verifier grade.
+
+The export includes successful and unsuccessful attempts on both kept and unselected tasks. Conversation availability is not a training-quality acceptance claim; no additional conversation quality filter is applied to this export. Historical attempts remain in v2.
+
+Token totals sum calls across an attempt and are not the context window.
+
+Measurements cover 999 tasks, including kept and unselected tasks, with each attempt counted once.
+
+## Evaluation settings
+
+| Model | Agent | Thinking | Attempts/task | Temperature | Top-p | Top-k | Context tokens | Intended output tokens/call | Other sampling |
+| --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
+| Qwen/Qwen3.8-27B | Terminus-2 | On | 2 | 1.0 | 0.95 | 20 | 32,768 | 8,192 | min_p=0; presence_penalty=0; repetition_penalty=1 |
+
+Context includes prompt and response for one model call. The output budget includes thinking tokens; neither is a whole-attempt token total.
+
+**Time limits:** agent 600 seconds for 1,972 attempts and 900 seconds for 26, set by task; verifier 120 seconds for all attempts.
+
+**Output-limit evidence:** All 1,998 saved configurations contain an explicit 8,192-token request cap. HTTP payloads and server enforcement were not audited.
+
+## Measurements
+
+### Qwen/Qwen3.8-27B
+
+| Measurement | Attempts measured | Median | 90th percentile |
+| --- | ---: | ---: | ---: |
+| Agent turns | 1,998 | 5 | 9 |
+| Input tokens per attempt | 1,998 | 13,384 | 57,469 |
+| Output tokens per attempt | 1,998 | 2,774 | 12,053 |
+| Largest recorded step, prompt plus output tokens | 1,998 | 4,604 | 13,212 |
+| Agent runtime (seconds) | 1,998 | 61.36 | 316.90 |
+| Verifier runtime (seconds) | 1,992 | 8.45 | 9.49 |
+| Whole-attempt runtime (seconds) | 1,998 | 81.23 | 337.99 |
+
+Statistics use available measurements only. Input/output totals sum calls and can include repeated context. Recorded ATIF steps can aggregate multiple requests, so their largest prompt-plus-output count does not establish a per-call context limit. Missing summarization counts are not zero. Commands are counted as submitted, not as successful executions.
+
+Of 1,998 attempts, 1,924 ended with the agent marking the task complete, 67 at the agent time limit, 6 with an environment error and 1 after repeated output truncation.
+
+Machine-readable measurements remain in `data/summary.json`, `data/per_task.parquet` and `data/per_attempt.parquet`.
+
+## 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)
+
+![Output, summarization and recorded context](plots/all-models-tokens-by-outcome.png)
+
+Output per recorded agent step is separated by verifier outcome. Summarizations are counted per trace, including zero when recorded. Peak recorded context is shown per trace with and without summarization; unknown counts stay separate.

fineenvs_v3/trace_stats/data/per_attempt.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcc95ea05c6c436f1bd138b745e3cd52819e68255d5d6cf4ed18c8e1e3556d85
+size 329215

fineenvs_v3/trace_stats/data/per_task.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa00a24c337ee61013b3e5b365c1274a373fff01c142d7f68256b2f171a64909
+size 321315

fineenvs_v3/trace_stats/data/provenance.json

@@ -0,0 +1,51 @@
+{
+ "deduplication_key": [
+  "model",
+  "task_id",
+  "attempt_id"
+ ],
+ "sources": [
+  {
+   "file": "traces/qwen38-thinking-0.parquet",
+   "rows": 500,
+   "unique_rows_added": 500,
+   "sha256": "242845880981c2fc49dd5bcfe69f14125accafb718fd8b56729244701745e38e"
+  },
+  {
+   "file": "traces/qwen38-thinking-1.parquet",
+   "rows": 500,
+   "unique_rows_added": 500,
+   "sha256": "8d53d553ef5738e2afc4cf93cb6764c3efdddeb160dc72ee084b8e7b97c11cdd"
+  },
+  {
+   "file": "traces/qwen38-thinking-2.parquet",
+   "rows": 500,
+   "unique_rows_added": 500,
+   "sha256": "25a8e8a28044186da3e48f48b3a01c0ed7ced658d53f1f4fe1aaabb75050ec84"
+  },
+  {
+   "file": "traces/qwen38-thinking-3.parquet",
+   "rows": 498,
+   "unique_rows_added": 498,
+   "sha256": "46ff0ceb08ce8d2e4f1bdfbe22f5157e1f3ec32ac73e58e77d3b859442794478"
+  }
+ ],
+ "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"
+  },
+  "qwen3.8-27b": {
+   "trace_distributions": "plots/qwen3.8-27b-trace-distributions.png",
+   "trace_outcomes": "plots/qwen3.8-27b-trace-outcomes.png",
+   "trace_commands": "plots/qwen3.8-27b-trace-commands.png"
+  }
+ },
+ "metric_implementation_sha256": "ffa2ef63f6c84b193cd50503ece64f6819abc82b894c041ae9c7fb601ef33fe1",
+ "outcome_implementation_sha256": "570e8bdb6930588cb87b9f3728518cae549fd20da595808a2ecef7fed75dddc0",
+ "plotting": {
+  "implementation_sha256": "3e3bdc00f8b2100de4f96bdb48ce6fd87eff518a6bb380f9b07c3da352ad7db6"
+ },
+ "outcome_policy": "exclusive-outcomes-v1"
+}

fineenvs_v3/trace_stats/data/summary.json

@@ -0,0 +1,1436 @@
+{
+ "scope": "All attempts in the published trace Parquets of this folder.",
+ "created_at": "2026-10-27T22:19:29.919982+00:00",
+ "input_rows": 1998,
+ "duplicate_records_removed": 0,
+ "unique_attempts": 1998,
+ "unique_tasks": 999,
+ "missing_trajectories": 0,
+ "null_verifier_rewards": 6,
+ "all_attempts": {
+  "trajectories": 1998,
+  "graded": 1998,
+  "solved": 897,
+  "turns": {
+   "n": 1998,
+   "mean": 5.594094094094094,
+   "median": 5,
+   "p90": 9,
+   "p95": 11,
+   "max": 20
+  },
+  "input_tokens": {
+   "n": 1998,
+   "mean": 24375.874374374373,
+   "median": 13384,
+   "p90": 57469,
+   "p95": 87635,
+   "max": 254708
+  },
+  "input_tokens_total": 48702997,
+  "output_tokens": {
+   "n": 1998,
+   "mean": 4956.957457457457,
+   "median": 2774,
+   "p90": 12053,
+   "p95": 17458,
+   "max": 36812
+  },
+  "output_tokens_total": 9904001,
+  "peak_context_tokens": {
+   "n": 1998,
+   "mean": 6639.290290290291,
+   "median": 4604,
+   "p90": 13212,
+   "p95": 17950,
+   "max": 46457
+  },
+  "peak_context_fraction": {
+   "n": 1998,
+   "mean": 0.20261506012848787,
+   "median": 0.1405029296875,
+   "p90": 0.4031982421875,
+   "p95": 0.54779052734375,
+   "max": 1.417755126953125
+  },
+  "trajectories_above_90_percent_context": 13,
+  "latency_seconds": {
+   "model_call": {
+    "n": 11165,
+    "mean": 13.874932032895632,
+    "median": 7.449318170547485,
+    "p90": 32.661933183670044,
+    "p95": 48.729466915130615,
+    "max": 136.15840339660645
+   },
+   "tool_call": null,
+   "agent": {
+    "n": 1998,
+    "mean": 123.11313489439439,
+    "median": 61.362441,
+    "p90": 316.901775,
+    "p95": 483.252097,
+    "max": 603.230676
+   },
+   "verifier": {
+    "n": 1992,
+    "mean": 8.583377089859438,
+    "median": 8.445156,
+    "p90": 9.493342,
+    "p95": 10.229738,
+    "max": 23.125993
+   },
+   "trial": {
+    "n": 1998,
+    "mean": 143.24902637437438,
+    "median": 81.229523,
+    "p90": 337.991541,
+    "p95": 503.594045,
+    "max": 764.450748
+   }
+  },
+  "tools": {
+   "bash_command": {
+    "calls": 13590,
+    "calls_per_trajectory": 6.801801801801802,
+    "failed": null,
+    "failure_rate": null
+   },
+   "mark_task_complete": {
+    "calls": 3995,
+    "calls_per_trajectory": 1.9994994994994995,
+    "failed": null,
+    "failure_rate": null
+   }
+  },
+  "tool_steps": 11022,
+  "tool_steps_without_output": 0,
+  "tool_steps_with_error_text": 789,
+  "tool_steps_with_error_text_rate": 0.07158410451823626,
+  "terminations": {
+   "environment_error": 6,
+   "repeated_output_truncation": 1,
+   "task_complete": 1924,
+   "task_timeout": 67
+  },
+  "command_distribution": {
+   "[assignment]": 0.00031638930937078456,
+   "[unknown]": 0.0018553259204475904,
+   "apt": 1.6145177435500014e-05,
+   "apt-cache": 1.668335001668335e-05,
+   "apt-get": 0.00014091602292706624,
+   "awk": 0.0006099950206940828,
+   "cat": 0.16885301963055835,
+   "cd": 0.0737891846309199,
+   "cp": 5.98000598000598e-05,
+   "cut": 0.0005676251419285476,
+   "df": 3.0860900426117815e-05,
+   "diff": 3.707411114818522e-05,
+   "echo": 0.14505262385744908,
+   "export": 3.33667000333667e-05,
+   "file": 0.0014380270123773597,
+   "find": 0.007022134159026405,
+   "free": 0.00016538070388901857,
+   "grep": 0.008289016329971937,
+   "head": 0.09729647621978928,
+   "hexdump": 5.005005005005005e-05,
+   "jobs": 1.82000182000182e-05,
+   "kill": 9.443405669820763e-06,
+   "ldconfig": 2.86000286000286e-05,
+   "ln": 1.668335001668335e-05,
+   "ls": 0.07504878078790124,
+   "md5sum": 1.7875017875017873e-05,
+   "mkdir": 0.08327001484715718,
+   "mv": 0.00022720720788353642,
+   "nproc": 8.780910903552414e-05,
+   "od": 0.000552732932907225,
+   "pip": 0.0026554174009669777,
+   "printf": 0.0012706301079194095,
+   "ps": 2.275002275002275e-05,
+   "pwd": 0.0008553844306650322,
+   "python": 0.28992040946265074,
+   "python-m:pip": 0.00013485707930152374,
+   "q": 2.275002275002275e-05,
+   "rm": 9.753110598521226e-05,
+   "sed": 0.0014005408739936256,
+   "sha256sum": 1.0214295928581642e-05,
+   "sleep": 7.563583553366441e-05,
+   "sort": 0.000415621090434352,
+   "sqlite3": 0.0007315636400603722,
+   "strings": 2.7679194345861013e-05,
+   "tail": 0.0042530803760999405,
+   "tee": 0.00011560473516995256,
+   "timeout": 0.00032037847485169666,
+   "tr": 0.00027212846024711533,
+   "true": 0.00042095319873097645,
+   "uniq": 0.00011836723793245531,
+   "wc": 0.030980963419687303,
+   "which": 0.0003818396332624189,
+   "xargs": 9.175842509175844e-05,
+   "xxd": 0.00048650046213399037
+  },
+  "output_cap_events": 0,
+  "oom_recoveries": 32,
+  "trajectories_with_oom_recovery": 26,
+  "malformed_tool_calls": 81,
+  "trajectories_with_malformed_tool_calls": 76,
+  "errors": {
+   "model_server": {
+    "trajectories": 0,
+    "types": {}
+   },
+   "environment": {
+    "trajectories": 6,
+    "types": {
+     "TmuxBatchProtocolError": 6
+    }
+   },
+   "verifier": {
+    "trajectories": 0,
+    "types": {}
+   },
+   "other": {
+    "trajectories": 0,
+    "types": {}
+   }
+  }
+ },
+ "models": {
+  "qwen3.8-27b": {
+   "trajectories": 1998,
+   "graded": 1998,
+   "solved": 897,
+   "turns": {
+    "n": 1998,
+    "mean": 5.594094094094094,
+    "median": 5,
+    "p90": 9,
+    "p95": 11,
+    "max": 20
+   },
+   "input_tokens": {
+    "n": 1998,
+    "mean": 24375.874374374373,
+    "median": 13384,
+    "p90": 57469,
+    "p95": 87635,
+    "max": 254708
+   },
+   "input_tokens_total": 48702997,
+   "output_tokens": {
+    "n": 1998,
+    "mean": 4956.957457457457,
+    "median": 2774,
+    "p90": 12053,
+    "p95": 17458,
+    "max": 36812
+   },
+   "output_tokens_total": 9904001,
+   "peak_context_tokens": {
+    "n": 1998,
+    "mean": 6639.290290290291,
+    "median": 4604,
+    "p90": 13212,
+    "p95": 17950,
+    "max": 46457
+   },
+   "peak_context_fraction": {
+    "n": 1998,
+    "mean": 0.20261506012848787,
+    "median": 0.1405029296875,
+    "p90": 0.4031982421875,
+    "p95": 0.54779052734375,
+    "max": 1.417755126953125
+   },
+   "trajectories_above_90_percent_context": 13,
+   "latency_seconds": {
+    "model_call": {
+     "n": 11165,
+     "mean": 13.874932032895632,
+     "median": 7.449318170547485,
+     "p90": 32.661933183670044,
+     "p95": 48.729466915130615,
+     "max": 136.15840339660645
+    },
+    "tool_call": null,
+    "agent": {
+     "n": 1998,
+     "mean": 123.11313489439439,
+     "median": 61.362441,
+     "p90": 316.901775,
+     "p95": 483.252097,
+     "max": 603.230676
+    },
+    "verifier": {
+     "n": 1992,
+     "mean": 8.583377089859438,
+     "median": 8.445156,
+     "p90": 9.493342,
+     "p95": 10.229738,
+     "max": 23.125993
+    },
+    "trial": {
+     "n": 1998,
+     "mean": 143.24902637437438,
+     "median": 81.229523,
+     "p90": 337.991541,
+     "p95": 503.594045,
+     "max": 764.450748
+    }
+   },
+   "tools": {
+    "bash_command": {
+     "calls": 13590,
+     "calls_per_trajectory": 6.801801801801802,
+     "failed": null,
+     "failure_rate": null
+    },
+    "mark_task_complete": {
+     "calls": 3995,
+     "calls_per_trajectory": 1.9994994994994995,
+     "failed": null,
+     "failure_rate": null
+    }
+   },
+   "tool_steps": 11022,
+   "tool_steps_without_output": 0,
+   "tool_steps_with_error_text": 789,
+   "tool_steps_with_error_text_rate": 0.07158410451823626,
+   "terminations": {
+    "environment_error": 6,
+    "repeated_output_truncation": 1,
+    "task_complete": 1924,
+    "task_timeout": 67
+   },
+   "command_distribution": {
+    "[assignment]": 0.00031638930937078456,
+    "[unknown]": 0.0018553259204475904,
+    "apt": 1.6145177435500014e-05,
+    "apt-cache": 1.668335001668335e-05,
+    "apt-get": 0.00014091602292706624,
+    "awk": 0.0006099950206940828,
+    "cat": 0.16885301963055835,
+    "cd": 0.0737891846309199,
+    "cp": 5.98000598000598e-05,
+    "cut": 0.0005676251419285476,
+    "df": 3.0860900426117815e-05,
+    "diff": 3.707411114818522e-05,
+    "echo": 0.14505262385744908,
+    "export": 3.33667000333667e-05,
+    "file": 0.0014380270123773597,
+    "find": 0.007022134159026405,
+    "free": 0.00016538070388901857,
+    "grep": 0.008289016329971937,
+    "head": 0.09729647621978928,
+    "hexdump": 5.005005005005005e-05,
+    "jobs": 1.82000182000182e-05,
+    "kill": 9.443405669820763e-06,
+    "ldconfig": 2.86000286000286e-05,
+    "ln": 1.668335001668335e-05,
+    "ls": 0.07504878078790124,
+    "md5sum": 1.7875017875017873e-05,
+    "mkdir": 0.08327001484715718,
+    "mv": 0.00022720720788353642,
+    "nproc": 8.780910903552414e-05,
+    "od": 0.000552732932907225,
+    "pip": 0.0026554174009669777,
+    "printf": 0.0012706301079194095,
+    "ps": 2.275002275002275e-05,
+    "pwd": 0.0008553844306650322,
+    "python": 0.28992040946265074,
+    "python-m:pip": 0.00013485707930152374,
+    "q": 2.275002275002275e-05,
+    "rm": 9.753110598521226e-05,
+    "sed": 0.0014005408739936256,
+    "sha256sum": 1.0214295928581642e-05,
+    "sleep": 7.563583553366441e-05,
+    "sort": 0.000415621090434352,
+    "sqlite3": 0.0007315636400603722,
+    "strings": 2.7679194345861013e-05,
+    "tail": 0.0042530803760999405,
+    "tee": 0.00011560473516995256,
+    "timeout": 0.00032037847485169666,
+    "tr": 0.00027212846024711533,
+    "true": 0.00042095319873097645,
+    "uniq": 0.00011836723793245531,
+    "wc": 0.030980963419687303,
+    "which": 0.0003818396332624189,
+    "xargs": 9.175842509175844e-05,
+    "xxd": 0.00048650046213399037
+   },
+   "output_cap_events": 0,
+   "oom_recoveries": 32,
+   "trajectories_with_oom_recovery": 26,
+   "malformed_tool_calls": 81,
+   "trajectories_with_malformed_tool_calls": 76,
+   "errors": {
+    "model_server": {
+     "trajectories": 0,
+     "types": {}
+    },
+    "environment": {
+     "trajectories": 6,
+     "types": {
+      "TmuxBatchProtocolError": 6
+     }
+    },
+    "verifier": {
+     "trajectories": 0,
+     "types": {}
+    },
+    "other": {
+     "trajectories": 0,
+     "types": {}
+    }
+   }
+  }
+ },
+ "model_ids": {
+  "qwen3.8-27b": "Qwen/Qwen3.8-27B"
+ },
+ "by_outcome": {
+  "all-models": {
+   "Successful": {
+    "trajectories": 897,
+    "graded": 897,
+    "solved": 897,
+    "turns": {
+     "n": 897,
+     "mean": 5.0891861761426975,
+     "median": 5,
+     "p90": 8,
+     "p95": 9,
+     "max": 17
+    },
+    "input_tokens": {
+     "n": 897,
+     "mean": 18500.198439241918,
+     "median": 11669,
+     "p90": 38119,
+     "p95": 59484,
+     "max": 182558
+    },
+    "input_tokens_total": 16594678,
+    "output_tokens": {
+     "n": 897,
+     "mean": 3743.3322185061315,
+     "median": 2342,
+     "p90": 7682,
+     "p95": 11924,
+     "max": 31535
+    },
+    "output_tokens_total": 3357769,
+    "peak_context_tokens": {
+     "n": 897,
+     "mean": 5532.711259754738,
+     "median": 4042,
+     "p90": 9805,
+     "p95": 13359,
+     "max": 44721
+    },
+    "peak_context_fraction": {
+     "n": 897,
+     "mean": 0.1688449481126324,
+     "median": 0.12335205078125,
+     "p90": 0.299224853515625,
+     "p95": 0.407684326171875,
+     "max": 1.364776611328125
+    },
+    "trajectories_above_90_percent_context": 4,
+    "latency_seconds": {
+     "model_call": {
+      "n": 4565,
+      "mean": 11.62472008598595,
+      "median": 6.588855981826782,
+      "p90": 25.89209794998169,
+      "p95": 36.78495526313782,
+      "max": 136.15840339660645
+     },
+     "tool_call": null,
+     "agent": {
+      "n": 897,
+      "mean": 91.71262544147157,
+      "median": 50.702327,
+      "p90": 215.311156,
+      "p95": 313.335338,
+      "max": 600.002441
+     },
+     "verifier": {
+      "n": 897,
+      "mean": 8.353372635451505,
+      "median": 8.196922,
+      "p90": 9.1396,
+      "p95": 9.989307,
+      "max": 13.603097
+     },
+     "trial": {
+      "n": 897,
+      "mean": 111.1160799386845,
+      "median": 70.354049,
+      "p90": 241.670764,
+      "p95": 337.991541,
+      "max": 622.690654
+     }
+    },
+    "tools": {
+     "bash_command": {
+      "calls": 5356,
+      "calls_per_trajectory": 5.971014492753623,
+      "failed": null,
+      "failure_rate": null
+     },
+     "mark_task_complete": {
+      "calls": 1843,
+      "calls_per_trajectory": 2.0546265328874025,
+      "failed": null,
+      "failure_rate": null
+     }
+    },
+    "tool_steps": 4529,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 277,
+    "tool_steps_with_error_text_rate": 0.06116140428350629,
+    "terminations": {
+     "task_complete": 895,
+     "task_timeout": 2
+    },
+    "command_distribution": {
+     "[assignment]": 0.0002638205832160352,
+     "[unknown]": 0.0025258866665300887,
+     "apt-get": 0.00014864362690449647,
+     "awk": 0.0009300223874166,
+     "cat": 0.16639222417188196,
+     "cd": 0.07088799168134116,
+     "cut": 0.0008171921854786305,
+     "echo": 0.15338617742188823,
+     "file": 0.0009195038371239744,
+     "find": 0.003158666441078867,
+     "free": 0.00023105995726996546,
+     "grep": 0.00555813418297017,
+     "head": 0.09365225021259684,
+     "hexdump": 0.00011148272017837236,
+     "kill": 2.103447550535327e-05,
+     "ls": 0.07222808125161077,
+     "md5sum": 3.9815257206561555e-05,
+     "mkdir": 0.09175125880862643,
+     "mv": 0.0001238696890870804,
+     "nproc": 0.0001753185971807793,
+     "od": 0.00039996235648409557,
+     "pip": 0.0026428655351690856,
+     "printf": 0.0009926043106821135,
+     "pwd": 0.0009075546657127203,
+     "python": 0.294897339267602,
+     "python-m:pip": 0.00013935340022296544,
+     "rm": 8.563165462976426e-05,
+     "sed": 0.0006917645713632336,
+     "sort": 0.0005946601594398659,
+     "sqlite3": 0.0012316906867369194,
+     "tail": 0.0032614105112455066,
+     "tee": 6.967670011148272e-05,
+     "timeout": 7.465360726230291e-05,
+     "tr": 0.00021569482817119866,
+     "true": 0.0001981915025393286,
+     "uniq": 0.00023831713340219957,
+     "wc": 0.029253355552369395,
+     "which": 0.0003788955195604812,
+     "xargs": 0.00011148272017837236,
+     "xxd": 0.00029246116202637945
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 19,
+    "trajectories_with_oom_recovery": 14,
+    "malformed_tool_calls": 34,
+    "trajectories_with_malformed_tool_calls": 31,
+    "errors": {
+     "model_server": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "environment": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "verifier": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "other": {
+      "trajectories": 0,
+      "types": {}
+     }
+    }
+   },
+   "Unsuccessful (graded)": {
+    "trajectories": 1095,
+    "graded": 1095,
+    "solved": 0,
+    "turns": {
+     "n": 1095,
+     "mean": 6.005479452054795,
+     "median": 5,
+     "p90": 10,
+     "p95": 12,
+     "max": 20
+    },
+    "input_tokens": {
+     "n": 1095,
+     "mean": 29178.440182648403,
+     "median": 15295,
+     "p90": 74839,
+     "p95": 103680,
+     "max": 254708
+    },
+    "input_tokens_total": 31950392,
+    "output_tokens": {
+     "n": 1095,
+     "mean": 5950.064840182648,
+     "median": 3298,
+     "p90": 15397,
+     "p95": 19761,
+     "max": 36812
+    },
+    "output_tokens_total": 6515321,
+    "peak_context_tokens": {
+     "n": 1095,
+     "mean": 7542.281278538813,
+     "median": 5237,
+     "p90": 15869,
+     "p95": 20202,
+     "max": 46457
+    },
+    "peak_context_fraction": {
+     "n": 1095,
+     "mean": 0.23017215815853312,
+     "median": 0.159820556640625,
+     "p90": 0.484283447265625,
+     "p95": 0.61651611328125,
+     "max": 1.417755126953125
+    },
+    "trajectories_above_90_percent_context": 9,
+    "latency_seconds": {
+     "model_call": {
+      "n": 6564,
+      "mean": 15.440107593625003,
+      "median": 8.150166988372803,
+      "p90": 37.457515716552734,
+      "p95": 53.57353377342224,
+      "max": 133.76567244529724
+     },
+     "tool_call": null,
+     "agent": {
+      "n": 1095,
+      "mean": 147.0902876958904,
+      "median": 74.513164,
+      "p90": 384.789619,
+      "p95": 600.002336,
+      "max": 603.230676
+     },
+     "verifier": {
+      "n": 1095,
+      "mean": 8.771791697716894,
+      "median": 8.649876,
+      "p90": 9.673962,
+      "p95": 10.407389,
+      "max": 23.125993
+     },
+     "trial": {
+      "n": 1095,
+      "mean": 167.3730998675799,
+      "median": 95.539557,
+      "p90": 411.032902,
+      "p95": 622.175391,
+      "max": 662.748356
+     }
+    },
+    "tools": {
+     "bash_command": {
+      "calls": 8169,
+      "calls_per_trajectory": 7.46027397260274,
+      "failed": null,
+      "failure_rate": null
+     },
+     "mark_task_complete": {
+      "calls": 2152,
+      "calls_per_trajectory": 1.965296803652968,
+      "failed": null,
+      "failure_rate": null
+     }
+    },
+    "tool_steps": 6463,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 503,
+    "tool_steps_with_error_text_rate": 0.07782763422559183,
+    "terminations": {
+     "repeated_output_truncation": 1,
+     "task_complete": 1029,
+     "task_timeout": 65
+    },
+    "command_distribution": {
+     "[assignment]": 0.00036118609769684415,
+     "[unknown]": 0.0013161834239057485,
+     "apt": 2.9459419649432907e-05,
+     "apt-cache": 3.0441400304414e-05,
+     "apt-get": 0.00013535788171227853,
+     "awk": 0.00035117805464300226,
+     "cat": 0.17127580860274574,
+     "cd": 0.07633004839411467,
+     "cp": 0.00010911462966257486,
+     "cut": 0.0003662955645652113,
+     "diff": 6.764755623203111e-05,
+     "echo": 0.13888601723488425,
+     "export": 6.0882800608828e-05,
+     "file": 0.001870669432721241,
+     "find": 0.010049106521979094,
+     "free": 5.617332022622528e-05,
+     "grep": 0.010376620944621051,
+     "head": 0.09987868814334719,
+     "ldconfig": 5.218525766470972e-05,
+     "ln": 3.0441400304414e-05,
+     "ls": 0.07709025820485459,
+     "mkdir": 0.07676203533799189,
+     "mv": 0.0003131040093517759,
+     "od": 0.0006809079143218277,
+     "pip": 0.002344178885210082,
+     "printf": 0.0015053451040558212,
+     "ps": 4.1511000415110007e-05,
+     "pwd": 0.0008173347555474197,
+     "python": 0.28633616307395615,
+     "python-m:pip": 0.00013191273465246067,
+     "q": 4.1511000415110007e-05,
+     "rm": 0.00010781329274479959,
+     "sed": 0.001988829082855199,
+     "sha256sum": 1.8637592023110614e-05,
+     "sleep": 0.00012140509699949159,
+     "sort": 0.00027123358508700956,
+     "sqlite3": 0.00032587909300238065,
+     "strings": 5.0505050505050505e-05,
+     "tail": 0.004601383647875423,
+     "tee": 0.00015386142545165772,
+     "timeout": 0.0004671158246465944,
+     "tr": 0.0003198487695928503,
+     "true": 0.0006057412906727975,
+     "uniq": 2.0755500207555003e-05,
+     "wc": 0.03215526092691683,
+     "which": 0.00036973925847556463,
+     "xargs": 7.6103500761035e-05,
+     "xxd": 0.0006481189598228771
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 13,
+    "trajectories_with_oom_recovery": 12,
+    "malformed_tool_calls": 47,
+    "trajectories_with_malformed_tool_calls": 45,
+    "errors": {
+     "model_server": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "environment": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "verifier": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "other": {
+      "trajectories": 0,
+      "types": {}
+     }
+    }
+   },
+   "No verifier score": {
+    "trajectories": 6,
+    "graded": 6,
+    "solved": 0,
+    "turns": {
+     "n": 6,
+     "mean": 6.0,
+     "median": 6,
+     "p90": 10,
+     "p95": 10,
+     "max": 10
+    },
+    "input_tokens": {
+     "n": 6,
+     "mean": 26321.166666666668,
+     "median": 20728,
+     "p90": 65396,
+     "p95": 65396,
+     "max": 65396
+    },
+    "input_tokens_total": 157927,
+    "output_tokens": {
+     "n": 6,
+     "mean": 5151.833333333333,
+     "median": 4485,
+     "p90": 10086,
+     "p95": 10086,
+     "max": 10086
+    },
+    "output_tokens_total": 30911,
+    "peak_context_tokens": {
+     "n": 6,
+     "mean": 7277.0,
+     "median": 7339,
+     "p90": 12862,
+     "p95": 12862,
+     "max": 12862
+    },
+    "peak_context_fraction": {
+     "n": 6,
+     "mean": 0.222076416015625,
+     "median": 0.223968505859375,
+     "p90": 0.39251708984375,
+     "p95": 0.39251708984375,
+     "max": 0.39251708984375
+    },
+    "trajectories_above_90_percent_context": 0,
+    "latency_seconds": {
+     "model_call": {
+      "n": 36,
+      "mean": 13.830630838871002,
+      "median": 7.759422779083252,
+      "p90": 30.498255968093872,
+      "p95": 55.57064437866211,
+      "max": 66.0119206905365
+     },
+     "tool_call": null,
+     "agent": {
+      "n": 6,
+      "mean": 441.6589118333333,
+      "median": 384.296132,
+      "p90": 585.389038,
+      "p95": 585.389038,
+      "max": 585.389038
+     },
+     "verifier": null,
+     "trial": {
+      "n": 6,
+      "mean": 544.4811060000001,
+      "median": 496.66978,
+      "p90": 764.450748,
+      "p95": 764.450748,
+      "max": 764.450748
+     }
+    },
+    "tools": {
+     "bash_command": {
+      "calls": 65,
+      "calls_per_trajectory": 10.833333333333334,
+      "failed": null,
+      "failure_rate": null
+     }
+    },
+    "tool_steps": 30,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 9,
+    "tool_steps_with_error_text_rate": 0.3,
+    "terminations": {
+     "environment_error": 6
+    },
+    "command_distribution": {
+     "cat": 0.09458295327860546,
+     "cd": 0.04380989380989381,
+     "df": 0.010276679841897233,
+     "echo": 0.024592074592074593,
+     "find": 0.03218810175331915,
+     "free": 0.010276679841897233,
+     "grep": 0.035568055133272526,
+     "head": 0.17085458824589259,
+     "jobs": 0.006060606060606061,
+     "ls": 0.12417373286938506,
+     "mkdir": 0.0030303030303030303,
+     "nproc": 0.0030303030303030303,
+     "pip": 0.061332950463385244,
+     "python": 0.19999436955958697,
+     "sleep": 0.0030303030303030303,
+     "tail": 0.08894237807281284,
+     "timeout": 0.010276679841897233,
+     "wc": 0.07494904451426192,
+     "which": 0.0030303030303030303
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 0,
+    "trajectories_with_oom_recovery": 0,
+    "malformed_tool_calls": 0,
+    "trajectories_with_malformed_tool_calls": 0,
+    "errors": {
+     "model_server": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "environment": {
+      "trajectories": 6,
+      "types": {
+       "TmuxBatchProtocolError": 6
+      }
+     },
+     "verifier": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "other": {
+      "trajectories": 0,
+      "types": {}
+     }
+    }
+   }
+  },
+  "qwen3.8-27b": {
+   "Successful": {
+    "trajectories": 897,
+    "graded": 897,
+    "solved": 897,
+    "turns": {
+     "n": 897,
+     "mean": 5.0891861761426975,
+     "median": 5,
+     "p90": 8,
+     "p95": 9,
+     "max": 17
+    },
+    "input_tokens": {
+     "n": 897,
+     "mean": 18500.198439241918,
+     "median": 11669,
+     "p90": 38119,
+     "p95": 59484,
+     "max": 182558
+    },
+    "input_tokens_total": 16594678,
+    "output_tokens": {
+     "n": 897,
+     "mean": 3743.3322185061315,
+     "median": 2342,
+     "p90": 7682,
+     "p95": 11924,
+     "max": 31535
+    },
+    "output_tokens_total": 3357769,
+    "peak_context_tokens": {
+     "n": 897,
+     "mean": 5532.711259754738,
+     "median": 4042,
+     "p90": 9805,
+     "p95": 13359,
+     "max": 44721
+    },
+    "peak_context_fraction": {
+     "n": 897,
+     "mean": 0.1688449481126324,
+     "median": 0.12335205078125,
+     "p90": 0.299224853515625,
+     "p95": 0.407684326171875,
+     "max": 1.364776611328125
+    },
+    "trajectories_above_90_percent_context": 4,
+    "latency_seconds": {
+     "model_call": {
+      "n": 4565,
+      "mean": 11.62472008598595,
+      "median": 6.588855981826782,
+      "p90": 25.89209794998169,
+      "p95": 36.78495526313782,
+      "max": 136.15840339660645
+     },
+     "tool_call": null,
+     "agent": {
+      "n": 897,
+      "mean": 91.71262544147157,
+      "median": 50.702327,
+      "p90": 215.311156,
+      "p95": 313.335338,
+      "max": 600.002441
+     },
+     "verifier": {
+      "n": 897,
+      "mean": 8.353372635451505,
+      "median": 8.196922,
+      "p90": 9.1396,
+      "p95": 9.989307,
+      "max": 13.603097
+     },
+     "trial": {
+      "n": 897,
+      "mean": 111.1160799386845,
+      "median": 70.354049,
+      "p90": 241.670764,
+      "p95": 337.991541,
+      "max": 622.690654
+     }
+    },
+    "tools": {
+     "bash_command": {
+      "calls": 5356,
+      "calls_per_trajectory": 5.971014492753623,
+      "failed": null,
+      "failure_rate": null
+     },
+     "mark_task_complete": {
+      "calls": 1843,
+      "calls_per_trajectory": 2.0546265328874025,
+      "failed": null,
+      "failure_rate": null
+     }
+    },
+    "tool_steps": 4529,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 277,
+    "tool_steps_with_error_text_rate": 0.06116140428350629,
+    "terminations": {
+     "task_complete": 895,
+     "task_timeout": 2
+    },
+    "command_distribution": {
+     "[assignment]": 0.0002638205832160352,
+     "[unknown]": 0.0025258866665300887,
+     "apt-get": 0.00014864362690449647,
+     "awk": 0.0009300223874166,
+     "cat": 0.16639222417188196,
+     "cd": 0.07088799168134116,
+     "cut": 0.0008171921854786305,
+     "echo": 0.15338617742188823,
+     "file": 0.0009195038371239744,
+     "find": 0.003158666441078867,
+     "free": 0.00023105995726996546,
+     "grep": 0.00555813418297017,
+     "head": 0.09365225021259684,
+     "hexdump": 0.00011148272017837236,
+     "kill": 2.103447550535327e-05,
+     "ls": 0.07222808125161077,
+     "md5sum": 3.9815257206561555e-05,
+     "mkdir": 0.09175125880862643,
+     "mv": 0.0001238696890870804,
+     "nproc": 0.0001753185971807793,
+     "od": 0.00039996235648409557,
+     "pip": 0.0026428655351690856,
+     "printf": 0.0009926043106821135,
+     "pwd": 0.0009075546657127203,
+     "python": 0.294897339267602,
+     "python-m:pip": 0.00013935340022296544,
+     "rm": 8.563165462976426e-05,
+     "sed": 0.0006917645713632336,
+     "sort": 0.0005946601594398659,
+     "sqlite3": 0.0012316906867369194,
+     "tail": 0.0032614105112455066,
+     "tee": 6.967670011148272e-05,
+     "timeout": 7.465360726230291e-05,
+     "tr": 0.00021569482817119866,
+     "true": 0.0001981915025393286,
+     "uniq": 0.00023831713340219957,
+     "wc": 0.029253355552369395,
+     "which": 0.0003788955195604812,
+     "xargs": 0.00011148272017837236,
+     "xxd": 0.00029246116202637945
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 19,
+    "trajectories_with_oom_recovery": 14,
+    "malformed_tool_calls": 34,
+    "trajectories_with_malformed_tool_calls": 31,
+    "errors": {
+     "model_server": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "environment": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "verifier": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "other": {
+      "trajectories": 0,
+      "types": {}
+     }
+    }
+   },
+   "Unsuccessful (graded)": {
+    "trajectories": 1095,
+    "graded": 1095,
+    "solved": 0,
+    "turns": {
+     "n": 1095,
+     "mean": 6.005479452054795,
+     "median": 5,
+     "p90": 10,
+     "p95": 12,
+     "max": 20
+    },
+    "input_tokens": {
+     "n": 1095,
+     "mean": 29178.440182648403,
+     "median": 15295,
+     "p90": 74839,
+     "p95": 103680,
+     "max": 254708
+    },
+    "input_tokens_total": 31950392,
+    "output_tokens": {
+     "n": 1095,
+     "mean": 5950.064840182648,
+     "median": 3298,
+     "p90": 15397,
+     "p95": 19761,
+     "max": 36812
+    },
+    "output_tokens_total": 6515321,
+    "peak_context_tokens": {
+     "n": 1095,
+     "mean": 7542.281278538813,
+     "median": 5237,
+     "p90": 15869,
+     "p95": 20202,
+     "max": 46457
+    },
+    "peak_context_fraction": {
+     "n": 1095,
+     "mean": 0.23017215815853312,
+     "median": 0.159820556640625,
+     "p90": 0.484283447265625,
+     "p95": 0.61651611328125,
+     "max": 1.417755126953125
+    },
+    "trajectories_above_90_percent_context": 9,
+    "latency_seconds": {
+     "model_call": {
+      "n": 6564,
+      "mean": 15.440107593625003,
+      "median": 8.150166988372803,
+      "p90": 37.457515716552734,
+      "p95": 53.57353377342224,
+      "max": 133.76567244529724
+     },
+     "tool_call": null,
+     "agent": {
+      "n": 1095,
+      "mean": 147.0902876958904,
+      "median": 74.513164,
+      "p90": 384.789619,
+      "p95": 600.002336,
+      "max": 603.230676
+     },
+     "verifier": {
+      "n": 1095,
+      "mean": 8.771791697716894,
+      "median": 8.649876,
+      "p90": 9.673962,
+      "p95": 10.407389,
+      "max": 23.125993
+     },
+     "trial": {
+      "n": 1095,
+      "mean": 167.3730998675799,
+      "median": 95.539557,
+      "p90": 411.032902,
+      "p95": 622.175391,
+      "max": 662.748356
+     }
+    },
+    "tools": {
+     "bash_command": {
+      "calls": 8169,
+      "calls_per_trajectory": 7.46027397260274,
+      "failed": null,
+      "failure_rate": null
+     },
+     "mark_task_complete": {
+      "calls": 2152,
+      "calls_per_trajectory": 1.965296803652968,
+      "failed": null,
+      "failure_rate": null
+     }
+    },
+    "tool_steps": 6463,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 503,
+    "tool_steps_with_error_text_rate": 0.07782763422559183,
+    "terminations": {
+     "repeated_output_truncation": 1,
+     "task_complete": 1029,
+     "task_timeout": 65
+    },
+    "command_distribution": {
+     "[assignment]": 0.00036118609769684415,
+     "[unknown]": 0.0013161834239057485,
+     "apt": 2.9459419649432907e-05,
+     "apt-cache": 3.0441400304414e-05,
+     "apt-get": 0.00013535788171227853,
+     "awk": 0.00035117805464300226,
+     "cat": 0.17127580860274574,
+     "cd": 0.07633004839411467,
+     "cp": 0.00010911462966257486,
+     "cut": 0.0003662955645652113,
+     "diff": 6.764755623203111e-05,
+     "echo": 0.13888601723488425,
+     "export": 6.0882800608828e-05,
+     "file": 0.001870669432721241,
+     "find": 0.010049106521979094,
+     "free": 5.617332022622528e-05,
+     "grep": 0.010376620944621051,
+     "head": 0.09987868814334719,
+     "ldconfig": 5.218525766470972e-05,
+     "ln": 3.0441400304414e-05,
+     "ls": 0.07709025820485459,
+     "mkdir": 0.07676203533799189,
+     "mv": 0.0003131040093517759,
+     "od": 0.0006809079143218277,
+     "pip": 0.002344178885210082,
+     "printf": 0.0015053451040558212,
+     "ps": 4.1511000415110007e-05,
+     "pwd": 0.0008173347555474197,
+     "python": 0.28633616307395615,
+     "python-m:pip": 0.00013191273465246067,
+     "q": 4.1511000415110007e-05,
+     "rm": 0.00010781329274479959,
+     "sed": 0.001988829082855199,
+     "sha256sum": 1.8637592023110614e-05,
+     "sleep": 0.00012140509699949159,
+     "sort": 0.00027123358508700956,
+     "sqlite3": 0.00032587909300238065,
+     "strings": 5.0505050505050505e-05,
+     "tail": 0.004601383647875423,
+     "tee": 0.00015386142545165772,
+     "timeout": 0.0004671158246465944,
+     "tr": 0.0003198487695928503,
+     "true": 0.0006057412906727975,
+     "uniq": 2.0755500207555003e-05,
+     "wc": 0.03215526092691683,
+     "which": 0.00036973925847556463,
+     "xargs": 7.6103500761035e-05,
+     "xxd": 0.0006481189598228771
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 13,
+    "trajectories_with_oom_recovery": 12,
+    "malformed_tool_calls": 47,
+    "trajectories_with_malformed_tool_calls": 45,
+    "errors": {
+     "model_server": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "environment": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "verifier": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "other": {
+      "trajectories": 0,
+      "types": {}
+     }
+    }
+   },
+   "No verifier score": {
+    "trajectories": 6,
+    "graded": 6,
+    "solved": 0,
+    "turns": {
+     "n": 6,
+     "mean": 6.0,
+     "median": 6,
+     "p90": 10,
+     "p95": 10,
+     "max": 10
+    },
+    "input_tokens": {
+     "n": 6,
+     "mean": 26321.166666666668,
+     "median": 20728,
+     "p90": 65396,
+     "p95": 65396,
+     "max": 65396
+    },
+    "input_tokens_total": 157927,
+    "output_tokens": {
+     "n": 6,
+     "mean": 5151.833333333333,
+     "median": 4485,
+     "p90": 10086,
+     "p95": 10086,
+     "max": 10086
+    },
+    "output_tokens_total": 30911,
+    "peak_context_tokens": {
+     "n": 6,
+     "mean": 7277.0,
+     "median": 7339,
+     "p90": 12862,
+     "p95": 12862,
+     "max": 12862
+    },
+    "peak_context_fraction": {
+     "n": 6,
+     "mean": 0.222076416015625,
+     "median": 0.223968505859375,
+     "p90": 0.39251708984375,
+     "p95": 0.39251708984375,
+     "max": 0.39251708984375
+    },
+    "trajectories_above_90_percent_context": 0,
+    "latency_seconds": {
+     "model_call": {
+      "n": 36,
+      "mean": 13.830630838871002,
+      "median": 7.759422779083252,
+      "p90": 30.498255968093872,
+      "p95": 55.57064437866211,
+      "max": 66.0119206905365
+     },
+     "tool_call": null,
+     "agent": {
+      "n": 6,
+      "mean": 441.6589118333333,
+      "median": 384.296132,
+      "p90": 585.389038,
+      "p95": 585.389038,
+      "max": 585.389038
+     },
+     "verifier": null,
+     "trial": {
+      "n": 6,
+      "mean": 544.4811060000001,
+      "median": 496.66978,
+      "p90": 764.450748,
+      "p95": 764.450748,
+      "max": 764.450748
+     }
+    },
+    "tools": {
+     "bash_command": {
+      "calls": 65,
+      "calls_per_trajectory": 10.833333333333334,
+      "failed": null,
+      "failure_rate": null
+     }
+    },
+    "tool_steps": 30,
+    "tool_steps_without_output": 0,
+    "tool_steps_with_error_text": 9,
+    "tool_steps_with_error_text_rate": 0.3,
+    "terminations": {
+     "environment_error": 6
+    },
+    "command_distribution": {
+     "cat": 0.09458295327860546,
+     "cd": 0.04380989380989381,
+     "df": 0.010276679841897233,
+     "echo": 0.024592074592074593,
+     "find": 0.03218810175331915,
+     "free": 0.010276679841897233,
+     "grep": 0.035568055133272526,
+     "head": 0.17085458824589259,
+     "jobs": 0.006060606060606061,
+     "ls": 0.12417373286938506,
+     "mkdir": 0.0030303030303030303,
+     "nproc": 0.0030303030303030303,
+     "pip": 0.061332950463385244,
+     "python": 0.19999436955958697,
+     "sleep": 0.0030303030303030303,
+     "tail": 0.08894237807281284,
+     "timeout": 0.010276679841897233,
+     "wc": 0.07494904451426192,
+     "which": 0.0030303030303030303
+    },
+    "output_cap_events": 0,
+    "oom_recoveries": 0,
+    "trajectories_with_oom_recovery": 0,
+    "malformed_tool_calls": 0,
+    "trajectories_with_malformed_tool_calls": 0,
+    "errors": {
+     "model_server": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "environment": {
+      "trajectories": 6,
+      "types": {
+       "TmuxBatchProtocolError": 6
+      }
+     },
+     "verifier": {
+      "trajectories": 0,
+      "types": {}
+     },
+     "other": {
+      "trajectories": 0,
+      "types": {}
+     }
+    }
+   }
+  }
+ },
+ "outcome_policy": "exclusive-outcomes-v1",
+ "evaluation_settings": [
+  {
+   "agent": "terminus-2",
+   "agent_timeout": "task-specific; see saved task.toml and resolved trial timeouts",
+   "attempts": 2,
+   "context_tokens": 32768,
+   "contract_sha256": "9a7c288630f67227220fd23a62bc0978f4db02623b18c6e03d84d10fff09d937",
+   "hf_repo": "Qwen/Qwen3.8-27B",
+   "intended_output_tokens_per_call": 8192,
+   "model": "qwen38-27b",
+   "output_limit_status": "present in saved request kwargs; HTTP payload not audited",
+   "request_output_tokens": 8192,
+   "resolved_timeouts": {
+    "agent": 600.0,
+    "agent_setup": 360.0,
+    "environment_build": 600.0,
+    "steps": [],
+    "trial_attempt": null,
+    "trial_cleanup_grace": 30.0,
+    "verifier": 120.0
+   },
+   "sampling": {
+    "min_p": 0,
+    "presence_penalty": 0,
+    "repetition_penalty": 1.0,
+    "temperature": 1.0,
+    "top_k": 20,
+    "top_p": 0.95
+   },
+   "thinking": true,
+   "weights_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+   "attempts_covered": 1972
+  },
+  {
+   "agent": "terminus-2",
+   "agent_timeout": "task-specific; see saved task.toml and resolved trial timeouts",
+   "attempts": 2,
+   "context_tokens": 32768,
+   "contract_sha256": "9a7c288630f67227220fd23a62bc0978f4db02623b18c6e03d84d10fff09d937",
+   "hf_repo": "Qwen/Qwen3.8-27B",
+   "intended_output_tokens_per_call": 8192,
+   "model": "qwen38-27b",
+   "output_limit_status": "present in saved request kwargs; HTTP payload not audited",
+   "request_output_tokens": 8192,
+   "resolved_timeouts": {
+    "agent": 900.0,
+    "agent_setup": 360.0,
+    "environment_build": 600.0,
+    "steps": [],
+    "trial_attempt": null,
+    "trial_cleanup_grace": 30.0,
+    "verifier": 120.0
+   },
+   "sampling": {
+    "min_p": 0,
+    "presence_penalty": 0,
+    "repetition_penalty": 1.0,
+    "temperature": 1.0,
+    "top_k": 20,
+    "top_p": 0.95
+   },
+   "thinking": true,
+   "weights_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
+   "attempts_covered": 26
+  }
+ ]
+}

fineenvs_v3/trace_stats/data/tokens-by-outcome.json

@@ -0,0 +1,447 @@
+{
+  "scope": "All deduplicated published attempts, including kept and excluded tasks.",
+  "outcome_definition": "Raw verifier reward >= 1 is successful; lower finite scores are graded unsuccessful; missing/nonfinite scores are ungraded.",
+  "measurement_notes": "Step distributions weight each recorded agent step equally. ATIF step usage may aggregate multiple requests, so these are not guaranteed per-call measurements. Context means recorded prompt plus completion tokens. Attempt output includes reported auxiliary generation. Cumulative input counts repeated history again and is not unique rollout length. Missing measurements are excluded, never replaced by zero. Summary counts are recorded events, not a measure of their effectiveness.",
+  "models": {
+    "all-models": {
+      "Successful": {
+        "attempts": 897,
+        "metrics": {
+          "step_output_tokens": {
+            "n": 4563,
+            "mean": 734.5746219592373,
+            "median": 401,
+            "p90": 1589,
+            "p95": 2334,
+            "max": 16647
+          },
+          "step_prompt_tokens": {
+            "n": 4563,
+            "mean": 3630.703703703704,
+            "median": 2730,
+            "p90": 7314,
+            "p95": 9712,
+            "max": 32664
+          },
+          "step_context_tokens": {
+            "n": 4563,
+            "mean": 4365.278325662941,
+            "median": 3317,
+            "p90": 8402,
+            "p95": 10996,
+            "max": 44721
+          },
+          "output_tokens": {
+            "n": 897,
+            "mean": 3743.3322185061315,
+            "median": 2342,
+            "p90": 7682,
+            "p95": 11924,
+            "max": 31535
+          },
+          "input_plus_output_tokens": {
+            "n": 897,
+            "mean": 22243.530657748048,
+            "median": 14138,
+            "p90": 45386,
+            "p95": 71056,
+            "max": 211373
+          },
+          "context_summarizations": {
+            "n": 897,
+            "mean": 0.0011148272017837235,
+            "median": 0,
+            "p90": 0,
+            "p95": 0,
+            "max": 1
+          }
+        },
+        "missing_measurements": {
+          "step_output_tokens": 0,
+          "step_prompt_tokens": 0,
+          "step_context_tokens": 0,
+          "output_tokens": 0,
+          "input_plus_output_tokens": 0,
+          "context_summarizations": 0
+        },
+        "attempts_with_summaries": 1
+      },
+      "Unsuccessful (graded)": {
+        "attempts": 1095,
+        "metrics": {
+          "step_output_tokens": {
+            "n": 6510,
+            "mean": 972.6861751152073,
+            "median": 497,
+            "p90": 2327,
+            "p95": 3377,
+            "max": 14345
+          },
+          "step_prompt_tokens": {
+            "n": 6510,
+            "mean": 4778.628110599078,
+            "median": 3340,
+            "p90": 10665,
+            "p95": 13797,
+            "max": 36841
+          },
+          "step_context_tokens": {
+            "n": 6510,
+            "mean": 5751.314285714286,
+            "median": 4126,
+            "p90": 12389,
+            "p95": 16081,
+            "max": 46457
+          },
+          "output_tokens": {
+            "n": 1095,
+            "mean": 5950.064840182648,
+            "median": 3298,
+            "p90": 15397,
+            "p95": 19761,
+            "max": 36812
+          },
+          "input_plus_output_tokens": {
+            "n": 1095,
+            "mean": 35128.50502283105,
+            "median": 18972,
+            "p90": 88084,
+            "p95": 119698,
+            "max": 279591
+          },
+          "context_summarizations": {
+            "n": 1095,
+            "mean": 0.014611872146118721,
+            "median": 0,
+            "p90": 0,
+            "p95": 0,
+            "max": 2
+          }
+        },
+        "missing_measurements": {
+          "step_output_tokens": 0,
+          "step_prompt_tokens": 0,
+          "step_context_tokens": 0,
+          "output_tokens": 0,
+          "input_plus_output_tokens": 0,
+          "context_summarizations": 0
+        },
+        "attempts_with_summaries": 10
+      },
+      "No verifier score": {
+        "attempts": 6,
+        "metrics": {
+          "step_output_tokens": {
+            "n": 30,
+            "mean": 930.7333333333333,
+            "median": 543,
+            "p90": 1901,
+            "p95": 3456,
+            "max": 4121
+          },
+          "step_prompt_tokens": {
+            "n": 30,
+            "mean": 3968.6666666666665,
+            "median": 3026,
+            "p90": 8526,
+            "p95": 10546,
+            "max": 11665
+          },
+          "step_context_tokens": {
+            "n": 30,
+            "mean": 4899.4,
+            "median": 3464,
+            "p90": 10305,
+            "p95": 12401,
+            "max": 12862
+          },
+          "output_tokens": {
+            "n": 6,
+            "mean": 5151.833333333333,
+            "median": 4485,
+            "p90": 10086,
+            "p95": 10086,
+            "max": 10086
+          },
+          "input_plus_output_tokens": {
+            "n": 6,
+            "mean": 31473.0,
+            "median": 27687,
+            "p90": 75482,
+            "p95": 75482,
+            "max": 75482
+          },
+          "context_summarizations": {
+            "n": 6,
+            "mean": 0.0,
+            "median": 0,
+            "p90": 0,
+            "p95": 0,
+            "max": 0
+          }
+        },
+        "missing_measurements": {
+          "step_output_tokens": 0,
+          "step_prompt_tokens": 0,
+          "step_context_tokens": 0,
+          "output_tokens": 0,
+          "input_plus_output_tokens": 0,
+          "context_summarizations": 0
+        },
+        "attempts_with_summaries": 0
+      },
+      "all_traces_summary_counts": {
+        "n": 1998,
+        "mean": 0.00850850850850851,
+        "median": 0,
+        "p90": 0,
+        "p95": 0,
+        "max": 2
+      },
+      "peak_context_by_summarization": {
+        "Without summarization": {
+          "distribution": {
+            "n": 1987,
+            "mean": 6538.163059889281,
+            "median": 4596,
+            "p90": 12809,
+            "p95": 17535,
+            "max": 46457
+          },
+          "missing": 0
+        },
+        "With summarization": {
+          "distribution": {
+            "n": 11,
+            "mean": 24906.545454545456,
+            "median": 24051,
+            "p90": 28407,
+            "p95": 32061,
+            "max": 32061
+          },
+          "missing": 0
+        }
+      }
+    },
+    "Qwen/Qwen3.8-27B": {
+      "Successful": {
+        "attempts": 897,
+        "metrics": {
+          "step_output_tokens": {
+            "n": 4563,
+            "mean": 734.5746219592373,
+            "median": 401,
+            "p90": 1589,
+            "p95": 2334,
+            "max": 16647
+          },
+          "step_prompt_tokens": {
+            "n": 4563,
+            "mean": 3630.703703703704,
+            "median": 2730,
+            "p90": 7314,
+            "p95": 9712,
+            "max": 32664
+          },
+          "step_context_tokens": {
+            "n": 4563,
+            "mean": 4365.278325662941,
+            "median": 3317,
+            "p90": 8402,
+            "p95": 10996,
+            "max": 44721
+          },
+          "output_tokens": {
+            "n": 897,
+            "mean": 3743.3322185061315,
+            "median": 2342,
+            "p90": 7682,
+            "p95": 11924,
+            "max": 31535
+          },
+          "input_plus_output_tokens": {
+            "n": 897,
+            "mean": 22243.530657748048,
+            "median": 14138,
+            "p90": 45386,
+            "p95": 71056,
+            "max": 211373
+          },
+          "context_summarizations": {
+            "n": 897,
+            "mean": 0.0011148272017837235,
+            "median": 0,
+            "p90": 0,
+            "p95": 0,
+            "max": 1
+          }
+        },
+        "missing_measurements": {
+          "step_output_tokens": 0,
+          "step_prompt_tokens": 0,
+          "step_context_tokens": 0,
+          "output_tokens": 0,
+          "input_plus_output_tokens": 0,
+          "context_summarizations": 0
+        },
+        "attempts_with_summaries": 1
+      },
+      "Unsuccessful (graded)": {
+        "attempts": 1095,
+        "metrics": {
+          "step_output_tokens": {
+            "n": 6510,
+            "mean": 972.6861751152073,
+            "median": 497,
+            "p90": 2327,
+            "p95": 3377,
+            "max": 14345
+          },
+          "step_prompt_tokens": {
+            "n": 6510,
+            "mean": 4778.628110599078,
+            "median": 3340,
+            "p90": 10665,
+            "p95": 13797,
+            "max": 36841
+          },
+          "step_context_tokens": {
+            "n": 6510,
+            "mean": 5751.314285714286,
+            "median": 4126,
+            "p90": 12389,
+            "p95": 16081,
+            "max": 46457
+          },
+          "output_tokens": {
+            "n": 1095,
+            "mean": 5950.064840182648,
+            "median": 3298,
+            "p90": 15397,
+            "p95": 19761,
+            "max": 36812
+          },
+          "input_plus_output_tokens": {
+            "n": 1095,
+            "mean": 35128.50502283105,
+            "median": 18972,
+            "p90": 88084,
+            "p95": 119698,
+            "max": 279591
+          },
+          "context_summarizations": {
+            "n": 1095,
+            "mean": 0.014611872146118721,
+            "median": 0,
+            "p90": 0,
+            "p95": 0,
+            "max": 2
+          }
+        },
+        "missing_measurements": {
+          "step_output_tokens": 0,
+          "step_prompt_tokens": 0,
+          "step_context_tokens": 0,
+          "output_tokens": 0,
+          "input_plus_output_tokens": 0,
+          "context_summarizations": 0
+        },
+        "attempts_with_summaries": 10
+      },
+      "No verifier score": {
+        "attempts": 6,
+        "metrics": {
+          "step_output_tokens": {
+            "n": 30,
+            "mean": 930.7333333333333,
+            "median": 543,
+            "p90": 1901,
+            "p95": 3456,
+            "max": 4121
+          },
+          "step_prompt_tokens": {
+            "n": 30,
+            "mean": 3968.6666666666665,
+            "median": 3026,
+            "p90": 8526,
+            "p95": 10546,
+            "max": 11665
+          },
+          "step_context_tokens": {
+            "n": 30,
+            "mean": 4899.4,
+            "median": 3464,
+            "p90": 10305,
+            "p95": 12401,
+            "max": 12862
+          },
+          "output_tokens": {
+            "n": 6,
+            "mean": 5151.833333333333,
+            "median": 4485,
+            "p90": 10086,
+            "p95": 10086,
+            "max": 10086
+          },
+          "input_plus_output_tokens": {
+            "n": 6,
+            "mean": 31473.0,
+            "median": 27687,
+            "p90": 75482,
+            "p95": 75482,
+            "max": 75482
+          },
+          "context_summarizations": {
+            "n": 6,
+            "mean": 0.0,
+            "median": 0,
+            "p90": 0,
+            "p95": 0,
+            "max": 0
+          }
+        },
+        "missing_measurements": {
+          "step_output_tokens": 0,
+          "step_prompt_tokens": 0,
+          "step_context_tokens": 0,
+          "output_tokens": 0,
+          "input_plus_output_tokens": 0,
+          "context_summarizations": 0
+        },
+        "attempts_with_summaries": 0
+      },
+      "all_traces_summary_counts": {
+        "n": 1998,
+        "mean": 0.00850850850850851,
+        "median": 0,
+        "p90": 0,
+        "p95": 0,
+        "max": 2
+      },
+      "peak_context_by_summarization": {
+        "Without summarization": {
+          "distribution": {
+            "n": 1987,
+            "mean": 6538.163059889281,
+            "median": 4596,
+            "p90": 12809,
+            "p95": 17535,
+            "max": 46457
+          },
+          "missing": 0
+        },
+        "With summarization": {
+          "distribution": {
+            "n": 11,
+            "mean": 24906.545454545456,
+            "median": 24051,
+            "p90": 28407,
+            "p95": 32061,
+            "max": 32061
+          },
+          "missing": 0
+        }
+      }
+    }
+  }
+}

fineenvs_v3/traces/manifest.json

@@ -0,0 +1,28 @@
+{
+  "scope": "All attempts of the v3 evaluation, successful and failed, on kept and still-unsolved tasks. Historical traces remain in fineenvs_v2.",
+  "model": "Qwen/Qwen3.8-27B",
+  "thinking": true,
+  "attempts": 1998,
+  "files": [
+    {
+      "file": "traces/qwen38-thinking-0.parquet",
+      "rows": 500,
+      "sha256": "242845880981c2fc49dd5bcfe69f14125accafb718fd8b56729244701745e38e"
+    },
+    {
+      "file": "traces/qwen38-thinking-1.parquet",
+      "rows": 500,
+      "sha256": "8d53d553ef5738e2afc4cf93cb6764c3efdddeb160dc72ee084b8e7b97c11cdd"
+    },
+    {
+      "file": "traces/qwen38-thinking-2.parquet",
+      "rows": 500,
+      "sha256": "25a8e8a28044186da3e48f48b3a01c0ed7ced658d53f1f4fe1aaabb75050ec84"
+    },
+    {
+      "file": "traces/qwen38-thinking-3.parquet",
+      "rows": 498,
+      "sha256": "46ff0ceb08ce8d2e4f1bdfbe22f5157e1f3ec32ac73e58e77d3b859442794478"
+    }
+  ]
+}

fineenvs_v3/traces/qwen38-thinking-0.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:242845880981c2fc49dd5bcfe69f14125accafb718fd8b56729244701745e38e
+size 2986010

fineenvs_v3/traces/qwen38-thinking-1.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d53d553ef5738e2afc4cf93cb6764c3efdddeb160dc72ee084b8e7b97c11cdd
+size 2636135

fineenvs_v3/traces/qwen38-thinking-2.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25a8e8a28044186da3e48f48b3a01c0ed7ced658d53f1f4fe1aaabb75050ec84
+size 2701641

fineenvs_v3/traces/qwen38-thinking-3.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46ff0ceb08ce8d2e4f1bdfbe22f5157e1f3ec32ac73e58e77d3b859442794478
+size 2851811