Comprehension accuracy (Δ)

These are the committed bytes rendered as readable JSON. Expanding this audit detail does not change the measurement’s current status.

{
    "construct": "<ACTION>, extra-retries(<n>) | <ACTION>, total-attempts(<n>) - a count-basis qualifier on a repeatable action; both forms state a ceiling, not a requirement to exhaust it",
    "metric": "comprehension_accuracy_delta",
    "seed": 20260920,
    "comparator": {
        "kind": "complete-careful-english-v1",
        "description": "Every compact form is compared with its complete registered careful-English meaning: for extra-retries(n), 'Make at most one initial execution of ACTION and, if success is not established, at most n additional executions: at most n+1 executions altogether.'; for total-attempts(n), 'Make at most n executions of ACTION altogether, including the first execution.' The ambiguous bare count phrase is absent from the scalar."
    },
    "study_purpose": "claim_test",
    "study_scope": "FRESH-INPUT replication of the DISPUTED original 97726167 (-55.355 pp; disputed, 1 disagreement, 0 confirmations; two quantized local readers). 72 fresh ACTION worlds x 2 strata = 144 real items (the source's 64 = 32 x 2 readers), 9 worlds per maximum 1..8, the source's hidden-intent pairing (extra-retries(m-1) and total-attempts(m) share the action and the maximum m), its four-option schema {max, max-1, max+1, 'the maximum is not specified'} and its VERBATIM consequence question, inherited deliberately so the row is comparable. Fresh actions across the ten domain families the prediction names, fresh case ids, 0 actions reused, 0 shared 8-grams from the marked arm; 68 shared 8-grams are the declared comparator's own wording and 6 are the inherited question. 16 fresh construct-free planted-position controls. READER POPULATION CHANGED, disclosed before spend: ONE hosted reader (deepseek-flash minimal), panel_neff 1; the source's two local readers are not matched. Filed unchanged.",
    "items_sha256": "c07f87f3b43e6d429b68d7308039b2ec6e11a43bec8b15a3fc86bc557c1b7af3",
    "items_url": "https://x0.at/Ti56.json",
    "models": [
        "deepseek-flash-minimal"
    ],
    "admissibility": {
        "kind": "ainglish.panel.admissibility.v1",
        "per_reader_calibration": true,
        "max_absent_cells": 4,
        "max_off_option_cells": 0,
        "max_transport_fault_cells": 4,
        "max_truncated_cells": 0
    },
    "readers": [
        {
            "name": "deepseek-flash-minimal",
            "provider": "openai-compatible",
            "model": "deepseek-flash",
            "api": "openai",
            "base_url": "https://api.deepseek.com/v1",
            "model_digest": null,
            "digest_source": "provider-opaque",
            "instrument_preparation": {
                "entry_point": "prepare_reader_instruments",
                "binding": "provider-opaque"
            },
            "answer_protocol": "opaque-choice-v1",
            "max_tokens": 32768,
            "timeout_s": 600,
            "temperature": null,
            "seed": "provider-default",
            "top_p": "provider-default",
            "top_k": "provider-default",
            "num_ctx": "provider-default",
            "reasoning_effort": "minimal"
        }
    ],
    "instrument_preparation": {
        "entry_point": "prepare_reader_instruments",
        "binding": [
            {
                "reader": "deepseek-flash-minimal",
                "digest_source": "provider-opaque"
            }
        ]
    },
    "item_counts": {
        "real": 144,
        "calibration": 16
    },
    "interval_kind": "bootstrap_items",
    "interval_estimator": {
        "kind": "ainglish.panel.bootstrap-items-attestation.v1",
        "algorithm": "sha256-counter-modulo-v1",
        "draws": 2000,
        "sampling_unit": "item",
        "quantiles": [
            "0.025",
            "0.975"
        ],
        "items_index_sha256": "e844100961f6936c80728759be71fd559bced2856d1500303cdffaade2c5113c"
    },
    "settlement_strata": [
        {
            "id": "extra-retries",
            "weight": 1
        },
        {
            "id": "total-attempts",
            "weight": 1
        }
    ],
    "settlement_item_field": "settlement_stratum",
    "settlement_rule": "manifest-weighted arms and value; every stratum load-bearing",
    "calibration": {
        "planted_arm": "ainglish",
        "min_gap": 0.5,
        "min_recovered": 0.875,
        "rule": "headroom-relative-v1",
        "ordering": "calibration-first",
        "arm_exposure": "both-arms-per-reader-item",
        "cells": 32
    },
    "difficulty": {
        "annotated": false
    },
    "harness": "ainglish-panel/0.2.58",
    "transport": {
        "deepseek-flash-minimal": {
            "max_tokens": 32768,
            "timeout_s": 600,
            "temperature": null,
            "seed": "provider-default",
            "top_p": "provider-default",
            "top_k": "provider-default",
            "num_ctx": "provider-default",
            "reasoning_effort": "minimal"
        }
    },
    "concurrency": {
        "max_in_flight": 4,
        "per_reader_max_in_flight": {
            "deepseek-flash-minimal": 4
        },
        "result_order": "deterministic-plan-order",
        "calibration_barrier": true,
        "automatic_retries": false
    },
    "transport_faults": {
        "total": 0,
        "retried": false,
        "per_cell": []
    },
    "transport_truncations": {
        "total": 0,
        "per_reader_cell": [],
        "by_cell": {
            "english": 0,
            "ainglish": 0
        },
        "imbalanced_across_cells": false
    },
    "protocol": "panel.py counterbalanced real arms + both-arms-per-reader-item planted-effect calibration gate"
}