{
  "schema": "s2s-bench-v1",
  "exhibit": "the-quantization-ladder",
  "dataset": "quantization-ladder-2026-10-14",
  "generated_utc": "2026-10-17T03:45:01Z",
  "licence": "CC BY 4.0",
  "attribution": "strata→signal research, research.strata2signal.com",
  "contact": "hello@strata2signal.com",
  "what_this_is": "The records behind the exhibit *The quantization ladder* (https://research.strata2signal.com/the-quantization-ladder/): Mistral NeMo 12B Instruct at five quantization levels (Q3_K_M, Q4_K_M, Q5_K_M, Q6_K, Q8_0), each run whole on one GeForce RTX 3090 at 300 W. Here: the pre-registration and its one amendment (public copies), the bench's own reading, every level's speed-ladder files (the reader's summary and the harness's raw file) and the smoke run's, the question step's figures, each level's arm record, probes, six-way report and six-way rows, the 2026-09-23 anchor the repeat was judged against (its doorman items withheld), the pull, stage and safety-watcher receipts, the one stop and the ladder's log. Paths keep the bench's own layout.",
  "run_id": "quantization-ladder-2026-10-14",
  "run_window_utc": {
    "smoke_utc": [
      "2026-10-15T05:31:26Z",
      "2026-10-15T05:32:20Z"
    ],
    "first_sitting_utc": [
      "2026-10-15T05:34:19Z",
      "2026-10-15T06:10:47Z"
    ],
    "second_sitting_utc": [
      "2026-10-15T10:33:13Z",
      "2026-10-15T11:01:54Z"
    ],
    "reading_written_utc": "2026-10-15T11:01:54Z",
    "note": "Two sittings, from the ladder's own log: the first ran Q4_K_M and Q8_0 whole and Q6_K's speed ladder, and stopped at Q6_K's question step; the second ran Q6_K's questions again from the start, then Q5_K_M and Q3_K_M. The smoke run checked the instrument before the first sitting."
  },
  "cut_by": {
    "tool": "tools/kit_sanitise.py",
    "how": "Every file below was COPIED from the bench's own record and then rewritten only by the named rules, applied in the order they are listed here, by one program. Nothing was edited by hand, nothing was reworded, and nothing was dropped from inside a file. `--check` re-derives every byte from the private record and refuses on a single byte of drift; `--fence` proves that no figure moved.",
    "rule_order": [
      "digest-withheld",
      "digest-withheld-ruled",
      "gpu-uuid-withheld",
      "serial-withheld",
      "seal-digest-withheld",
      "doorman-items-withheld",
      "port-withheld",
      "bus-address-withheld",
      "local-clock-phrase-withheld",
      "home-path-to-workshop",
      "lane-folder-to-card-key",
      "box-name-to-role-in-identifier",
      "box-name-to-role",
      "parallel-env-to-role",
      "operator-name-to-role"
    ],
    "files_cut": 60
  },
  "sanitisation_rules": {
    "digest-withheld": "the withholding rule every kit cut by rule on this shelf has carried since 2026-09-29 (UTC): a sha256 the record itself printed — whole, shortened to its first 8 or more hex characters, or written head…tail — of a file as it stood BEFORE a rule below rewrote it, or of an earlier version of such a file, becomes the literal `withheld-2026-09-29`. Beside the rewritten copy, such a digest would check a guess at what a rule replaced. Here that is the pre-registration's own digest as its FROZEN line prints it (the file's first 188 lines) and as the amendment cites it. The digests are derived by this program from the private originals, never listed by hand. A digest of a file no rule touched is left as the record printed it.",
    "digest-withheld-ruled": "added 2026-10-16 (UTC), under the house rule that a printed digest may not confirm a guess at what a rule on this shelf replaced (D-20260929-013): a digest of a file ANOTHER kit on this shelf publishes rewritten becomes the literal `withheld-2026-10-16` wherever this record prints it — whole, shortened to its first 8 or more hex characters, or written head…tail. One file: the Jev bench's runner, which Reading the answer's kit publishes rewritten and which the instrument records among the files it ran. The value is derived by this program from the file its specification names by revision and path, never listed by hand.",
    "gpu-uuid-withheld": "added 2026-10-05 (UTC), after publication: a GPU UUID — the identifier the driver prints for one physical board, `GPU-` and its 8-4-4-4-12 hex UUID (or a MIG instance's `MIG-` forms), whole or shortened to its first eight hex characters and any groups after them — becomes `<withheld: GPU UUID of board A>`, its prefix and any truncation mark with it, and so does a bare UUID straight after a GPU label (`gpu_uuid:`). The letter names the board inside this kit only: the kit's boards take A, B, C… in the order they first appear in its files, keyed by a UUID's first eight hex characters, so a shortened form takes its board's letter and a file that told two boards apart by their UUIDs still does. The harnesses recorded the board each run was pinned to in its rows, power samples and receipts; a UUID names one board and measures nothing. The rule matches the shape, never a value, which this program does not hold. Every other byte is unchanged.",
    "serial-withheld": "the value after a `device.serial:` or `ups.serial:` label, where the battery backup's own status readout prints the unit's serial number, becomes the literal `<withheld: hardware serial>`. Every raw speed-ladder file carries that readout, captured by the harness before and after each level; a serial names one physical unit and measures nothing. The rule matches the label and the shape of the value, never the value itself, which this program does not hold. Every other byte of the readout is unchanged.",
    "seal-digest-withheld": "A rule here cuts a span by its shape, never by a value it holds, and writes `withheld-2026-10-16` in its place. The two digests the amendment's stamp lines print (`sha256 of PREREG-QUANT.md's bytes followed by every byte above this line`): each hashes the private pre-registration and the private amendment as they stood, so beside these rewritten copies it would check a guess at what a rule replaced.",
    "doorman-items-withheld": "A rule here cuts a span by its shape, never by a value it holds, and writes `withheld-2026-10-16` in its place. The 2026-09-23 anchor's 48 doorman items (task `d`), the whole list, as a JSON string: each item is 16 hex characters of a hash of a private item's id and of its prompt, with that reading's verdict, and the doorman's set is never published. The count (`n`) and the score (`strict_correct`) beside the list are kept.",
    "port-withheld": "A rule here cuts a span by its shape, never by a value it holds, and writes `withheld-2026-10-16` in its place. The port the bench's own runtime listened on, after `127.0.0.1:` in a URL.",
    "bus-address-withheld": "A rule here cuts a span by its shape, never by a value it holds, and writes `withheld-2026-10-16` in its place. A card's PCI bus address (domain, bus, device and function, as the driver prints it, and the short bus-device-function form a note writes): the records tell the two boards apart by their withheld UUIDs' letters, and a bus address names a slot in one computer.",
    "local-clock-phrase-withheld": "A rule here cuts a span by its shape, never by a value it holds, and writes `withheld-2026-10-16` in its place. A phrase in the pre-registration's verbatim authority quote that names a time on an operator's own clock; this shelf writes times in UTC only.",
    "home-path-to-workshop": "an absolute path into a private home directory becomes the neutral root /workshop. The rest of the path and every byte around it are unchanged; the account's name is a path segment here and leaves with the root.",
    "lane-folder-to-card-key": "the reference card's lane folder in the cost-of-intelligence series, whose name carries a computer's private name, becomes the card key that series' kit publishes it under (`rtx-3090-reading-0`), in every path that cites it.",
    "box-name-to-role-in-identifier": "a private name at the start of a service unit's name, in a harness note about the bench computer's boot-time power-cap unit, becomes the role: `bench`.",
    "box-name-to-role": "the computer that holds the bench's card becomes the role it played, spelled as a name so that every sentence and identifier it stands in still reads as it did (the strict parser's version string among them, as the cost-of-intelligence kit prints it). This shelf publishes what a bench measured, never which box ran it.",
    "parallel-env-to-role": "the NAME of the runtime environment variable that sets how many requests the instance serves at once is generalised to the role it plays. ITS VALUE IS UNTOUCHED: it is one of this bench's own settings, and the value is what a reader needs.",
    "operator-name-to-role": "the operator named in the pre-registration's authority line, a recon note or a receipt becomes \"an operator\", which is also this shelf's own voice law for a published page."
  },
  "rule_firings_total": {
    "digest-withheld": 2,
    "digest-withheld-ruled": 6,
    "local-clock-phrase-withheld": 1,
    "lane-folder-to-card-key": 3,
    "box-name-to-role": 58,
    "operator-name-to-role": 7,
    "seal-digest-withheld": 2,
    "bus-address-withheld": 123,
    "doorman-items-withheld": 1,
    "gpu-uuid-withheld": 1938,
    "serial-withheld": 24,
    "port-withheld": 22,
    "box-name-to-role-in-identifier": 12,
    "parallel-env-to-role": 6,
    "home-path-to-workshop": 71
  },
  "what_was_not_touched": "Only what a rule rewrote, each rule counted per file in `provenance.json`: every GPU UUID to a board letter; the battery backup's serial to `<withheld: hardware serial>`; the amendment's two seal digests, the anchor's 48 doorman items, the runtime's port, the cards' bus addresses and one local-time phrase in the pre-registration's authority quote to `withheld-2026-10-16`; the home path to `/workshop`; the reference card's lane folder to its card key; the bench computer's and the operator's private names to the roles they played; one environment variable's name to its role; and the pre-registration's frozen digest to `withheld-2026-09-29`. No figure moved on the way out, and `index.json` carries the sha256 of every file it lists, as served.",
  "omitted": {
    "The doorman's set.": "No doorman (`.d.`) file is here, and the anchor's 48 doorman items are withheld: the set's items, questions and answers were never published. Its scores are in `reading.json`, `READING.md` and `results/tasks/figures.json`, as counts.",
    "The harness and the instrument.": "The measuring code is not here. The pre-registration names each program and its sha256; the speed ladder is the language bank this shelf's card pages publish, and the question instrument is the cost-of-intelligence series' (its frozen six-way set is in Reading the answer's kit, `kit/task_c.json`).",
    "The rest of the bench's tree.": "Only the files the page and its receipts cite are here: the halted step's moved-aside folder, the per-sample power and wall CSVs, the instance logs and the bench's own state files are not.",
    "Card identity.": "Every file names a card by its model and a board letter, never by a serial, a UUID or a bus address, except the safety watcher's 12 receipts (`receipts/stop-*.guard.json`), which name both boards `GPU-<redacted>`, as the watcher wrote them: its `core_temp_c_max_by_board` carries that key twice, and a JSON reader keeps only the second value (README.md says how to read both)."
  },
  "files": [
    {
      "file": "PREREG-QUANT.md",
      "role": "The pre-registration, frozen 2026-10-14T20:52:44Z before any scored call: the question, the five builds and their manifest digests, the instruments, the order, the eight predictions with their bars, the gates and what the page may say. A public copy cut by the rules in `provenance.json`; the digest its FROZEN line prints is withheld, because it hashes the private original.",
      "media_type": "text/markdown",
      "bytes": 16631,
      "sha256": "40607a11804285f6bcacca9e8f7c62233902173c32a83b8c3b8d1252e60553af",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld",
        "digest-withheld-ruled",
        "local-clock-phrase-withheld",
        "lane-folder-to-card-key",
        "box-name-to-role",
        "operator-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld": 1,
        "digest-withheld-ruled": 1,
        "local-clock-phrase-withheld": 1,
        "lane-folder-to-card-key": 2,
        "box-name-to-role": 19,
        "operator-name-to-role": 6
      }
    },
    {
      "file": "AMENDMENT-1.md",
      "role": "The pre-registration's one amendment, stamped 2026-10-14T22:15:53Z and 2026-10-15T02:18:44Z, before any scored call: the question step beside an idle second board (A1), the repeat gate when the repeat cannot be judged (A2), a path (A3), and the card picked by its link (A4). A public copy cut by the same rules; the digests its two stamp lines print are withheld.",
      "media_type": "text/markdown",
      "bytes": 6202,
      "sha256": "f232b1907c0b07ce2e4e4f5da99c4a097a321731e1cb37d38f52e2b23a1ae3c4",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld",
        "seal-digest-withheld",
        "bus-address-withheld",
        "box-name-to-role",
        "operator-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld": 1,
        "seal-digest-withheld": 2,
        "bus-address-withheld": 3,
        "box-name-to-role": 7,
        "operator-name-to-role": 1
      }
    },
    {
      "file": "READING.md",
      "role": "The reading, written by the bench's own reader at 2026-10-15T11:01Z after the last level: the speed table, the answers table, each prediction's verdict with the figures that decided it, and the one stop.",
      "media_type": "text/markdown",
      "bytes": 4209,
      "sha256": "a725aad9b730d6c901ddae860531a0f74710cf02681179a0d428366eb6437ea7",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "box-name-to-role"
      ],
      "strip_counts": {
        "box-name-to-role": 1
      }
    },
    {
      "file": "reading.json",
      "role": "The same reading as data: every figure in READING.md, per level, with each prediction's verdict and the inputs it was judged on.",
      "media_type": "application/json",
      "bytes": 8611,
      "sha256": "68384f0100d29843813fbedd928f066361ef6068f54518341ba26eda99554cdc",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "box-name-to-role"
      ],
      "strip_counts": {
        "box-name-to-role": 1
      }
    },
    {
      "file": "anchor/anchor-q4_K_M.json",
      "role": "The 2026-09-23 Q4_K_M reading the repeat (prediction 4) was judged against: its strict choice on each of the 108 six-way items, keyed by 16 hex characters of a hash of the item's id and of its prompt. Its 48 doorman items are withheld whole (the doorman's set is private); the counts beside them are kept.",
      "media_type": "application/json",
      "bytes": 10778,
      "sha256": "9b912475e4e8132862881fd01a2882ddf9a09d7eb96ac6f4bd581565a77ed64c",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "doorman-items-withheld",
        "lane-folder-to-card-key",
        "box-name-to-role"
      ],
      "strip_counts": {
        "doorman-items-withheld": 1,
        "lane-folder-to-card-key": 1,
        "box-name-to-role": 1
      }
    },
    {
      "file": "results/speed/READ-gpu-ql3090-24g-nemo-q3-k-m-control-3090-q3_K_M-ladder.json",
      "role": "One level's speed ladder as the bench's reader summarised it: per window, whether the model was whole on the card, the median writing speed, first token, board watts, joules per 1,000 tokens, run range, spread and gates.",
      "media_type": "application/json",
      "bytes": 4874,
      "sha256": "d7fd7723b9e554b529a45506bffbacf34c68fceb5f8bdb94d59bb8fe6a573cf7",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 8
      }
    },
    {
      "file": "results/speed/READ-gpu-ql3090-24g-nemo-q4-k-m-control-3090-q4_K_M-ladder.json",
      "role": "One level's speed ladder as the bench's reader summarised it: per window, whether the model was whole on the card, the median writing speed, first token, board watts, joules per 1,000 tokens, run range, spread and gates.",
      "media_type": "application/json",
      "bytes": 4872,
      "sha256": "3d6ee137352b62bbc5408b9706f8cf1ba228c0e3e97db9534edd46085b2294c5",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 8
      }
    },
    {
      "file": "results/speed/READ-gpu-ql3090-24g-nemo-q4-k-m-control-3090-q4_K_M-smoke-ladder.json",
      "role": "One level's speed ladder as the bench's reader summarised it: per window, whether the model was whole on the card, the median writing speed, first token, board watts, joules per 1,000 tokens, run range, spread and gates. This is the smoke run of 2026-10-15T05:31Z (Q4_K_M at 4,096 tokens only), which checked the instrument before the ladder and sits outside the registered order.",
      "media_type": "application/json",
      "bytes": 1012,
      "sha256": "922e0b5c7b121aba6448d480e6e3cd52b6a0122f0f633be43d8f07673e47af9c",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 1
      }
    },
    {
      "file": "results/speed/READ-gpu-ql3090-24g-nemo-q5-k-m-control-3090-q5_K_M-ladder.json",
      "role": "One level's speed ladder as the bench's reader summarised it: per window, whether the model was whole on the card, the median writing speed, first token, board watts, joules per 1,000 tokens, run range, spread and gates.",
      "media_type": "application/json",
      "bytes": 4876,
      "sha256": "d62c5e5b9ee83a3148d6e2a2ca6ed941b5c633d4f773199043fcd52e7e780dd8",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 8
      }
    },
    {
      "file": "results/speed/READ-gpu-ql3090-24g-nemo-q6-k-control-3090-q6_K-ladder.json",
      "role": "One level's speed ladder as the bench's reader summarised it: per window, whether the model was whole on the card, the median writing speed, first token, board watts, joules per 1,000 tokens, run range, spread and gates.",
      "media_type": "application/json",
      "bytes": 4862,
      "sha256": "44c7e2262226eb55315a7a1af37df554dc3108473c0281abcfe7ca10832f626a",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 8
      }
    },
    {
      "file": "results/speed/READ-gpu-ql3090-24g-nemo-q8-0-control-3090-q8_0-ladder.json",
      "role": "One level's speed ladder as the bench's reader summarised it: per window, whether the model was whole on the card, the median writing speed, first token, board watts, joules per 1,000 tokens, run range, spread and gates.",
      "media_type": "application/json",
      "bytes": 5177,
      "sha256": "6ffc0793eb4d5a3808b44b68a1038a0ef8b8785bd7715499dfa4a0da394cfdd0",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 9
      }
    },
    {
      "file": "results/speed/gpu-ql3090-24g-nemo-q3-k-m-control-3090-q3_K_M-ladder.json",
      "role": "One level's speed ladder as the harness wrote it: every scored run's timings and token counts, the card's power and memory samples, the posture reads, the fit verdict per window and the battery backup's readout before and after.",
      "media_type": "application/json",
      "bytes": 406594,
      "sha256": "af721e013666e286e9845543fc501496f5bb21f6154ccc2d11724824c91b479f",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "serial-withheld",
        "port-withheld",
        "bus-address-withheld",
        "box-name-to-role-in-identifier",
        "parallel-env-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 199,
        "serial-withheld": 4,
        "port-withheld": 1,
        "bus-address-withheld": 3,
        "box-name-to-role-in-identifier": 2,
        "parallel-env-to-role": 1
      }
    },
    {
      "file": "results/speed/gpu-ql3090-24g-nemo-q4-k-m-control-3090-q4_K_M-ladder.json",
      "role": "One level's speed ladder as the harness wrote it: every scored run's timings and token counts, the card's power and memory samples, the posture reads, the fit verdict per window and the battery backup's readout before and after.",
      "media_type": "application/json",
      "bytes": 409279,
      "sha256": "b1281566d719e48f833bf424096d6851040a81f697ae8984a679304247e996d2",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "serial-withheld",
        "port-withheld",
        "bus-address-withheld",
        "box-name-to-role-in-identifier",
        "parallel-env-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 199,
        "serial-withheld": 4,
        "port-withheld": 1,
        "bus-address-withheld": 3,
        "box-name-to-role-in-identifier": 2,
        "parallel-env-to-role": 1
      }
    },
    {
      "file": "results/speed/gpu-ql3090-24g-nemo-q4-k-m-control-3090-q4_K_M-smoke-ladder.json",
      "role": "One level's speed ladder as the harness wrote it: every scored run's timings and token counts, the card's power and memory samples, the posture reads, the fit verdict per window and the battery backup's readout before and after. This is the smoke run of 2026-10-15T05:31Z (Q4_K_M at 4,096 tokens only), which checked the instrument before the ladder and sits outside the registered order.",
      "media_type": "application/json",
      "bytes": 64115,
      "sha256": "4276d7f4919691e74028f19fe99491cff206ce9772686d732ee3e775398639d6",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "serial-withheld",
        "port-withheld",
        "bus-address-withheld",
        "box-name-to-role-in-identifier",
        "parallel-env-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 31,
        "serial-withheld": 4,
        "port-withheld": 1,
        "bus-address-withheld": 3,
        "box-name-to-role-in-identifier": 2,
        "parallel-env-to-role": 1
      }
    },
    {
      "file": "results/speed/gpu-ql3090-24g-nemo-q5-k-m-control-3090-q5_K_M-ladder.json",
      "role": "One level's speed ladder as the harness wrote it: every scored run's timings and token counts, the card's power and memory samples, the posture reads, the fit verdict per window and the battery backup's readout before and after.",
      "media_type": "application/json",
      "bytes": 411138,
      "sha256": "0fe6e11a13b2f289bd8243b9be1e8586301c2be3a6af3dcbdfe1576757d81544",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "serial-withheld",
        "port-withheld",
        "bus-address-withheld",
        "box-name-to-role-in-identifier",
        "parallel-env-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 199,
        "serial-withheld": 4,
        "port-withheld": 1,
        "bus-address-withheld": 3,
        "box-name-to-role-in-identifier": 2,
        "parallel-env-to-role": 1
      }
    },
    {
      "file": "results/speed/gpu-ql3090-24g-nemo-q6-k-control-3090-q6_K-ladder.json",
      "role": "One level's speed ladder as the harness wrote it: every scored run's timings and token counts, the card's power and memory samples, the posture reads, the fit verdict per window and the battery backup's readout before and after.",
      "media_type": "application/json",
      "bytes": 410155,
      "sha256": "79bac3339518e3bac6d6270fdce26022ba19d0f6bff335817026310c7011fadb",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "serial-withheld",
        "port-withheld",
        "bus-address-withheld",
        "box-name-to-role-in-identifier",
        "parallel-env-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 199,
        "serial-withheld": 4,
        "port-withheld": 1,
        "bus-address-withheld": 3,
        "box-name-to-role-in-identifier": 2,
        "parallel-env-to-role": 1
      }
    },
    {
      "file": "results/speed/gpu-ql3090-24g-nemo-q8-0-control-3090-q8_0-ladder.json",
      "role": "One level's speed ladder as the harness wrote it: every scored run's timings and token counts, the card's power and memory samples, the posture reads, the fit verdict per window and the battery backup's readout before and after.",
      "media_type": "application/json",
      "bytes": 362046,
      "sha256": "520200a5fe8c5d572579c643bf7313e5751133d4767219d40d696313bff1dbf2",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "serial-withheld",
        "port-withheld",
        "bus-address-withheld",
        "box-name-to-role-in-identifier",
        "parallel-env-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 179,
        "serial-withheld": 4,
        "port-withheld": 1,
        "bus-address-withheld": 3,
        "box-name-to-role-in-identifier": 2,
        "parallel-env-to-role": 1
      }
    },
    {
      "file": "results/tasks/figures.json",
      "role": "The question step's figures for every level: the six-way and the doorman's set scored strict and lenient, board and wall energy, timings, the probes and the fit on the named card. The doorman's set appears as counts only.",
      "media_type": "application/json",
      "bytes": 48105,
      "sha256": "dabef85cbba3611fee7e15f8a291285f11a5a2fea87f154396bfa2cf125d390d",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "box-name-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 20,
        "box-name-to-role": 11
      }
    },
    {
      "file": "results/tasks/q-q8_0/arm.json",
      "role": "One level's question step as the instrument wrote it: the model and its file header (parameter count, context length), the card's posture before and after, the fit and growth on the named card, the probes' summary and the step's start and end in UTC.",
      "media_type": "application/json",
      "bytes": 34315,
      "sha256": "7a7622211331ba93694a33b52a9764c84b1556cd3f556fe301e835354203ab7a",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld-ruled",
        "gpu-uuid-withheld",
        "port-withheld",
        "bus-address-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld-ruled": 1,
        "gpu-uuid-withheld": 49,
        "port-withheld": 1,
        "bus-address-withheld": 18,
        "home-path-to-workshop": 6,
        "box-name-to-role": 1
      }
    },
    {
      "file": "results/tasks/q-q8_0/probes.json",
      "role": "One level's probes: writing 256 tokens three times, reading a long prompt, and the first token on 12 six-way prompts, with the board's energy over each.",
      "media_type": "application/json",
      "bytes": 9131,
      "sha256": "33f1ae330e75f613b873eb38f036fb917621f35b0608069ef59165f620280f35",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 4
      }
    },
    {
      "file": "results/tasks/q-q8_0/q-q8_0.c.report.json",
      "role": "One level's six-way report as the instrument wrote it: the score, the energy window and the idle reading, and the instrument's settings.",
      "media_type": "application/json",
      "bytes": 5780,
      "sha256": "9a2a12f2893a75af3ee9ad52a595f9b131f6cc0b8ec4f4d033e2bb077f4b3ee1",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "port-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 13,
        "port-withheld": 1,
        "home-path-to-workshop": 1,
        "box-name-to-role": 2
      }
    },
    {
      "file": "results/tasks/q-q8_0/q-q8_0.c.jsonl",
      "role": "One level's six-way rows: one line per item (108), with the reply, the parse, the choice, the server's own timings and the card's posture.",
      "media_type": "application/x-ndjson",
      "bytes": 118617,
      "sha256": "d6447e9587cd1ab609fceab06f44b58ae530868ee439d12f6dd5748eeeab31de",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 108
      }
    },
    {
      "file": "results/tasks/q-q6_K/arm.json",
      "role": "One level's question step as the instrument wrote it: the model and its file header (parameter count, context length), the card's posture before and after, the fit and growth on the named card, the probes' summary and the step's start and end in UTC.",
      "media_type": "application/json",
      "bytes": 34020,
      "sha256": "11ec943e424179cc3a8083f65c2309d0c1fc7e2083e8128361602608877ad14e",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld-ruled",
        "gpu-uuid-withheld",
        "port-withheld",
        "bus-address-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld-ruled": 1,
        "gpu-uuid-withheld": 49,
        "port-withheld": 1,
        "bus-address-withheld": 18,
        "home-path-to-workshop": 6,
        "box-name-to-role": 1
      }
    },
    {
      "file": "results/tasks/q-q6_K/probes.json",
      "role": "One level's probes: writing 256 tokens three times, reading a long prompt, and the first token on 12 six-way prompts, with the board's energy over each.",
      "media_type": "application/json",
      "bytes": 8813,
      "sha256": "85aa133163e6d5d2a92896e53ca68f87590a66cb915a563551225b139db94506",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 4
      }
    },
    {
      "file": "results/tasks/q-q6_K/q-q6_K.c.report.json",
      "role": "One level's six-way report as the instrument wrote it: the score, the energy window and the idle reading, and the instrument's settings.",
      "media_type": "application/json",
      "bytes": 5836,
      "sha256": "af6445a4c3ea65d9f4469f50cce7cd6b2bcebe0055a3f59fe6d56ca3dc08de12",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "port-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 13,
        "port-withheld": 1,
        "home-path-to-workshop": 1,
        "box-name-to-role": 2
      }
    },
    {
      "file": "results/tasks/q-q6_K/q-q6_K.c.jsonl",
      "role": "One level's six-way rows: one line per item (108), with the reply, the parse, the choice, the server's own timings and the card's posture.",
      "media_type": "application/x-ndjson",
      "bytes": 118726,
      "sha256": "4e1e1e3b36064846478450b0d263c203a33beb431d2a9bb57b34095494abce15",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 108
      }
    },
    {
      "file": "results/tasks/q-q5_K_M/arm.json",
      "role": "One level's question step as the instrument wrote it: the model and its file header (parameter count, context length), the card's posture before and after, the fit and growth on the named card, the probes' summary and the step's start and end in UTC.",
      "media_type": "application/json",
      "bytes": 34016,
      "sha256": "125ae897da17f69427f70c0d7dd39cf15b5e450e2adb15e48ae829e6bc49d526",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld-ruled",
        "gpu-uuid-withheld",
        "port-withheld",
        "bus-address-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld-ruled": 1,
        "gpu-uuid-withheld": 49,
        "port-withheld": 1,
        "bus-address-withheld": 18,
        "home-path-to-workshop": 6,
        "box-name-to-role": 1
      }
    },
    {
      "file": "results/tasks/q-q5_K_M/probes.json",
      "role": "One level's probes: writing 256 tokens three times, reading a long prompt, and the first token on 12 six-way prompts, with the board's energy over each.",
      "media_type": "application/json",
      "bytes": 8812,
      "sha256": "df93475b56db169fc47de17a86db38dbff2b78bc0a254c9d097fd12146931d43",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 4
      }
    },
    {
      "file": "results/tasks/q-q5_K_M/q-q5_K_M.c.report.json",
      "role": "One level's six-way report as the instrument wrote it: the score, the energy window and the idle reading, and the instrument's settings.",
      "media_type": "application/json",
      "bytes": 5842,
      "sha256": "5cbd2f72c81996f7884e1cd87a7c85381af30d9a28e285895dabed76ffc3f697",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "port-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 13,
        "port-withheld": 1,
        "home-path-to-workshop": 1,
        "box-name-to-role": 2
      }
    },
    {
      "file": "results/tasks/q-q5_K_M/q-q5_K_M.c.jsonl",
      "role": "One level's six-way rows: one line per item (108), with the reply, the parse, the choice, the server's own timings and the card's posture.",
      "media_type": "application/x-ndjson",
      "bytes": 119346,
      "sha256": "78e96892a4815d10492543e5c6be7924956de8f618f57d9e0cf2efc63978487a",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 108
      }
    },
    {
      "file": "results/tasks/q-q4_K_M/arm.json",
      "role": "One level's question step as the instrument wrote it: the model and its file header (parameter count, context length), the card's posture before and after, the fit and growth on the named card, the probes' summary and the step's start and end in UTC.",
      "media_type": "application/json",
      "bytes": 34336,
      "sha256": "2470ebbbee5145a56fbc11c98cf9a85a273fec6840b6748059135b75e6da31b5",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld-ruled",
        "gpu-uuid-withheld",
        "port-withheld",
        "bus-address-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld-ruled": 1,
        "gpu-uuid-withheld": 49,
        "port-withheld": 1,
        "bus-address-withheld": 18,
        "home-path-to-workshop": 6,
        "box-name-to-role": 1
      }
    },
    {
      "file": "results/tasks/q-q4_K_M/probes.json",
      "role": "One level's probes: writing 256 tokens three times, reading a long prompt, and the first token on 12 six-way prompts, with the board's energy over each.",
      "media_type": "application/json",
      "bytes": 9133,
      "sha256": "d8196bcd0fa182e82fe19859555b79483a298a5458177d40d919f476517bdbd0",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 4
      }
    },
    {
      "file": "results/tasks/q-q4_K_M/q-q4_K_M.c.report.json",
      "role": "One level's six-way report as the instrument wrote it: the score, the energy window and the idle reading, and the instrument's settings.",
      "media_type": "application/json",
      "bytes": 5846,
      "sha256": "4b09fa2cba0b853cbf825cd3b51032db028e3961f9645fdf0b98ab166e7a5189",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "port-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 13,
        "port-withheld": 1,
        "home-path-to-workshop": 1,
        "box-name-to-role": 2
      }
    },
    {
      "file": "results/tasks/q-q4_K_M/q-q4_K_M.c.jsonl",
      "role": "One level's six-way rows: one line per item (108), with the reply, the parse, the choice, the server's own timings and the card's posture.",
      "media_type": "application/x-ndjson",
      "bytes": 119260,
      "sha256": "1baa5cf4398fb0079ed573f9788c737331fa18b48fe255f76fc843ae2673c173",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 108
      }
    },
    {
      "file": "results/tasks/q-q3_K_M/arm.json",
      "role": "One level's question step as the instrument wrote it: the model and its file header (parameter count, context length), the card's posture before and after, the fit and growth on the named card, the probes' summary and the step's start and end in UTC.",
      "media_type": "application/json",
      "bytes": 34055,
      "sha256": "a652b5365cf6c46bd36b9e0323726ebb0fc9cf50e926a4f745240da2c31fed63",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "digest-withheld-ruled",
        "gpu-uuid-withheld",
        "port-withheld",
        "bus-address-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "digest-withheld-ruled": 1,
        "gpu-uuid-withheld": 49,
        "port-withheld": 1,
        "bus-address-withheld": 18,
        "home-path-to-workshop": 6,
        "box-name-to-role": 1
      }
    },
    {
      "file": "results/tasks/q-q3_K_M/probes.json",
      "role": "One level's probes: writing 256 tokens three times, reading a long prompt, and the first token on 12 six-way prompts, with the board's energy over each.",
      "media_type": "application/json",
      "bytes": 8818,
      "sha256": "28abc242c2a282507ab08799613064eed3265ec0c363f9d6bd3a462267ff6b84",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 4
      }
    },
    {
      "file": "results/tasks/q-q3_K_M/q-q3_K_M.c.report.json",
      "role": "One level's six-way report as the instrument wrote it: the score, the energy window and the idle reading, and the instrument's settings.",
      "media_type": "application/json",
      "bytes": 5845,
      "sha256": "c4f543a69b4b60ba1ab4ec73d02e1d66a8716a4225de19bbb50134bb66b84570",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld",
        "port-withheld",
        "home-path-to-workshop",
        "box-name-to-role"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 13,
        "port-withheld": 1,
        "home-path-to-workshop": 1,
        "box-name-to-role": 2
      }
    },
    {
      "file": "results/tasks/q-q3_K_M/q-q3_K_M.c.jsonl",
      "role": "One level's six-way rows: one line per item (108), with the reply, the parse, the choice, the server's own timings and the card's posture.",
      "media_type": "application/x-ndjson",
      "bytes": 119350,
      "sha256": "fa57a6f72b8d576703ede6778145eb3344446e47c503f3eda2f0b18c8780bf42",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "gpu-uuid-withheld"
      ],
      "strip_counts": {
        "gpu-uuid-withheld": 108
      }
    },
    {
      "file": "receipts/pull-q3_K_M.json",
      "role": "A level's pull receipt: the tag, the pinned digest, the manifest's digest and size, whether the build was already in the bench's store, and the disk floor.",
      "media_type": "application/json",
      "bytes": 544,
      "sha256": "2408bd8f491c44d057c2e62ce54399958783b01851ea4a85ffce8d7b7f436287",
      "sha256_original": "2408bd8f491c44d057c2e62ce54399958783b01851ea4a85ffce8d7b7f436287",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    },
    {
      "file": "receipts/pull-q4_K_M.json",
      "role": "A level's pull receipt: the tag, the pinned digest, the manifest's digest and size, whether the build was already in the bench's store, and the disk floor.",
      "media_type": "application/json",
      "bytes": 499,
      "sha256": "eea7a328b5b678e4b6b7a3dd41a6e2a6b8bd6d0843d8bb89a39017bd33a2326e",
      "sha256_original": "eea7a328b5b678e4b6b7a3dd41a6e2a6b8bd6d0843d8bb89a39017bd33a2326e",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    },
    {
      "file": "receipts/pull-q5_K_M.json",
      "role": "A level's pull receipt: the tag, the pinned digest, the manifest's digest and size, whether the build was already in the bench's store, and the disk floor.",
      "media_type": "application/json",
      "bytes": 544,
      "sha256": "e4d20ad07dd8d1f3237fc335d241b978a0e6cf133099083cbc820ee53c501a58",
      "sha256_original": "e4d20ad07dd8d1f3237fc335d241b978a0e6cf133099083cbc820ee53c501a58",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    },
    {
      "file": "receipts/pull-q6_K.json",
      "role": "A level's pull receipt: the tag, the pinned digest, the manifest's digest and size, whether the build was already in the bench's store, and the disk floor.",
      "media_type": "application/json",
      "bytes": 497,
      "sha256": "d4a8387f9748c3bf3d4ce326e868fac5f0a295e4a8d08814220de345674eade1",
      "sha256_original": "d4a8387f9748c3bf3d4ce326e868fac5f0a295e4a8d08814220de345674eade1",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    },
    {
      "file": "receipts/pull-q8_0.json",
      "role": "A level's pull receipt: the tag, the pinned digest, the manifest's digest and size, whether the build was already in the bench's store, and the disk floor.",
      "media_type": "application/json",
      "bytes": 542,
      "sha256": "74a062cdaff856fb66a8902e95c318cb9d9d87c12e3343ed5c068ec60ad4b8b8",
      "sha256_original": "74a062cdaff856fb66a8902e95c318cb9d9d87c12e3343ed5c068ec60ad4b8b8",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    },
    {
      "file": "receipts/stage-20261015T051947Z.txt",
      "role": "A stage receipt, read before a sitting began: the card picked and its link, its posture, the runtime, the disk, and the battery backup's load.",
      "media_type": "text/plain",
      "bytes": 1239,
      "sha256": "0a1de370d82a62eb6d1c72e3447ea5f753828178358c22bb3fda90b445f50db4",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "bus-address-withheld",
        "box-name-to-role"
      ],
      "strip_counts": {
        "bus-address-withheld": 4,
        "box-name-to-role": 1
      }
    },
    {
      "file": "receipts/stage-20261015T053354Z.txt",
      "role": "A stage receipt, read before a sitting began: the card picked and its link, its posture, the runtime, the disk, and the battery backup's load.",
      "media_type": "text/plain",
      "bytes": 1243,
      "sha256": "02a7f6ea03c3643a745e54eddfb540e149915bf8aca9f0eb54863b7251d5c2fd",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "bus-address-withheld",
        "box-name-to-role"
      ],
      "strip_counts": {
        "bus-address-withheld": 4,
        "box-name-to-role": 1
      }
    },
    {
      "file": "receipts/stage-20261015T103251Z.txt",
      "role": "A stage receipt, read before a sitting began: the card picked and its link, its posture, the runtime, the disk, and the battery backup's load.",
      "media_type": "text/plain",
      "bytes": 1236,
      "sha256": "32d928f465e0a2aa41fc5eebed23146071e61e57ef5e227f2b36aeb7b7bc3cf3",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "bus-address-withheld",
        "box-name-to-role"
      ],
      "strip_counts": {
        "bus-address-withheld": 4,
        "box-name-to-role": 1
      }
    },
    {
      "file": "receipts/stop-smoke-q4_K_M-20261015T053125Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1466,
      "sha256": "fab0b078e3ad76d9eab7c9943b99c0fc36ed06e5fdb05bda2bdfc2fa8f382f8d",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "port-withheld",
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "port-withheld": 1,
        "home-path-to-workshop": 4
      }
    },
    {
      "file": "receipts/stop-speed-q3_K_M-20261015T105040Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1505,
      "sha256": "0a4596c7e093386986eb292fb3de973899e364807940eb988b42714ef2a155f7",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "port-withheld",
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "port-withheld": 1,
        "home-path-to-workshop": 4
      }
    },
    {
      "file": "receipts/stop-speed-q4_K_M-20261015T053420Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1505,
      "sha256": "b5f067e313874d143a66997d377b3f5865e1fb0da4fec0b62bfe99d85aa6515c",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "port-withheld",
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "port-withheld": 1,
        "home-path-to-workshop": 4
      }
    },
    {
      "file": "receipts/stop-speed-q5_K_M-20261015T103805Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1505,
      "sha256": "d0907e9196687ef7e0d782801234983638dd2fa0089981d07c4b8c5271519a03",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "port-withheld",
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "port-withheld": 1,
        "home-path-to-workshop": 4
      }
    },
    {
      "file": "receipts/stop-speed-q6_K-20261015T060056Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1499,
      "sha256": "75229ad53dbdbbc51bc2589e20fa13d308eeebfc0c87773aa5bbc339837f558d",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "port-withheld",
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "port-withheld": 1,
        "home-path-to-workshop": 4
      }
    },
    {
      "file": "receipts/stop-speed-q8_0-20261015T054759Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1499,
      "sha256": "e08b62fd168f963035f5c867e4347d0433e266f35a52a1b55c253e768436f409",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "port-withheld",
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "port-withheld": 1,
        "home-path-to-workshop": 4
      }
    },
    {
      "file": "receipts/stop-tasks-q3_K_M-20261015T105840Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1105,
      "sha256": "7e7ed27b7559e57959809301b954acb4d724a21bb8ca5c42d0a2bb2e099d718d",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "home-path-to-workshop": 2
      }
    },
    {
      "file": "receipts/stop-tasks-q4_K_M-20261015T054143Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1105,
      "sha256": "3e39be9bf793f42293df3d41dcc55596870821ba87c00b6a22a80091a5cc3423",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "home-path-to-workshop": 2
      }
    },
    {
      "file": "receipts/stop-tasks-q5_K_M-20261015T104619Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1105,
      "sha256": "5af9f61e6d928a4d45b80d1413aa0aa1dec6a9b6fcbfd3186575bfa2403d5f0b",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "home-path-to-workshop": 2
      }
    },
    {
      "file": "receipts/stop-tasks-q6_K-20261015T061017Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1102,
      "sha256": "d37b3704bd42be279a53b744be1fb4ff52968cdb1d551a362e79444103c7a9e0",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "home-path-to-workshop": 2
      }
    },
    {
      "file": "receipts/stop-tasks-q6_K-20261015T061017Z.jsonl",
      "role": "The one stop: the card watcher's line when it stopped Q6_K's first question step at 2026-10-15T06:10:40Z.",
      "media_type": "application/x-ndjson",
      "bytes": 132,
      "sha256": "5a675119d31ea8fe177cfbff4f8f0a4c086f7b65d34cb2f5ea5997551340b63b",
      "sha256_original": "5a675119d31ea8fe177cfbff4f8f0a4c086f7b65d34cb2f5ea5997551340b63b",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    },
    {
      "file": "receipts/stop-tasks-q6_K-20261015T103319Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1099,
      "sha256": "3e48cdc19ea5d1e11b437d485e57d72d955006348db5387c88c6f9f80d387d61",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "home-path-to-workshop": 2
      }
    },
    {
      "file": "receipts/stop-tasks-q8_0-20261015T055559Z.guard.json",
      "role": "The safety watcher's receipt for one step: the battery backup's highest load and each board's highest core temperature over the step, against the stop thresholds (80 % load, 83 °C).",
      "media_type": "application/json",
      "bytes": 1099,
      "sha256": "b4ab8c6c6aaad563b09fb4a1b7769324142c46e5a5a9a9270412de6732786b4a",
      "sha256_original": null,
      "identical_to_original": false,
      "stripped": [
        "home-path-to-workshop"
      ],
      "strip_counts": {
        "home-path-to-workshop": 2
      }
    },
    {
      "file": "logs/ladder.log",
      "role": "The ladder's own log: each level's start and end in UTC, the halt of 2026-10-15T06:10:47Z and the restart at 10:33:13Z.",
      "media_type": "text/plain",
      "bytes": 1337,
      "sha256": "a918f762d1d9f11e701fdacf2b361bc5a096f720b307478cf11e4ec7a1b6a041",
      "sha256_original": "a918f762d1d9f11e701fdacf2b361bc5a096f720b307478cf11e4ec7a1b6a041",
      "identical_to_original": true,
      "stripped": [],
      "strip_counts": {}
    }
  ]
}