{
  "schema": "outilsia.decision_quality.report.v1",
  "status": "completed",
  "status_detail": null,
  "protocol_version": "dragon-uncertainty.v1",
  "bridge_schema": "outilsia.dragon.agent-bridge.v1",
  "prompt_version": "outilsia.decision_quality.prompt.v1",
  "prompt_sha256": "84637504c86b4f714d1b3756bcc63c5bcae572a3d4e2f3c3c995e41d6b144a4d",
  "scaffolding": "observation_jsonl_only.v1",
  "response_format": "ollama_format_json+bridge_decision.v1",
  "model": {
    "model": "mistral:latest",
    "quantization": "Q4_K_M",
    "parameter_size": "7.2B",
    "family": "llama",
    "digest_prefix": "6577803aa9a0"
  },
  "runtime": "ollama",
  "ollama_version": "0.18.2",
  "app_version": "0.1.2",
  "build_id": "local-dev",
  "config": {
    "seeds": [
      0,
      1,
      2,
      3,
      4,
      5,
      6,
      7,
      8,
      9
    ],
    "move_timeout_seconds": 30,
    "max_rounds": 150,
    "max_decisions": 300,
    "max_attempts_per_decision": 3
  },
  "generation_options": {
    "num_ctx": 4096,
    "num_predict": 160,
    "seed": 42,
    "temperature": 0
  },
  "started_at_ms": 1791551371077,
  "finished_at_ms": 1791552416613,
  "summary": {
    "games": 10,
    "wins": 0,
    "win_rate": 0.0,
    "win_rate_ci95_low": 0.0,
    "win_rate_ci95_high": 0.2775,
    "treasure_pickups": 0,
    "treasure_rate": 0.0,
    "mean_rounds": 150.0,
    "mean_decisions": 177.8,
    "mean_hits": 0.1,
    "mean_bumps": 64.5,
    "known_wall_repeats": 634,
    "outcomes": {
      "timeout": 10
    },
    "llm_calls": 1778,
    "invalid_responses": 0,
    "invalid_rate": 0.0,
    "invalid_by_reason": {},
    "timed_out_calls": 0,
    "timeout_retries": 0,
    "timeouts_after_retry": 0,
    "timeout_rate": 0.0,
    "timeout_warning_threshold": 0.05,
    "measurement_perturbed": false,
    "measurement_warning": null,
    "forced_passes": 0,
    "dropped_probability_fields": 0,
    "mean_latency_ms": 583.0,
    "calibration": {
      "definition": "P(le déplacement annoncé ne heurte pas de mur) ; déplacements hors pass",
      "samples": 957,
      "brier_score": 0.5061,
      "brier_reference_constant_half": 0.25,
      "mean_confidence": 0.8492,
      "observed_move_success_rate": 0.326,
      "overconfidence": 0.5232,
      "method": "brier+reliability_10_bins+murphy_1973+ece(bootstrap_1000_mt19937_seed42)+wilson_per_bin.v2",
      "bins": 10,
      "min_samples_for_verdict": 30,
      "min_samples_per_bin": 5,
      "sample_sufficient": true,
      "observed_rate_ci95_low": 0.2971,
      "observed_rate_ci95_high": 0.3564,
      "reliability_bins": [
        {
          "lower": 0.5,
          "upper": 0.6,
          "samples": 55,
          "mean_confidence": 0.5,
          "observed_rate": 0.6545,
          "observed_rate_ci95_low": 0.5225,
          "observed_rate_ci95_high": 0.7664,
          "gap": -0.1545,
          "sufficient": true
        },
        {
          "lower": 0.8,
          "upper": 0.9,
          "samples": 296,
          "mean_confidence": 0.8,
          "observed_rate": 0.1858,
          "observed_rate_ci95_low": 0.1456,
          "observed_rate_ci95_high": 0.234,
          "gap": 0.6142,
          "sufficient": true
        },
        {
          "lower": 0.9,
          "upper": 1.0,
          "samples": 606,
          "mean_confidence": 0.905,
          "observed_rate": 0.3647,
          "observed_rate_ci95_low": 0.3273,
          "observed_rate_ci95_high": 0.4038,
          "gap": 0.5403,
          "sufficient": true
        }
      ],
      "ece": 0.541,
      "ece_ci95_low": 0.5116,
      "ece_ci95_high": 0.5692,
      "mce": 0.6142,
      "murphy": {
        "reliability": 0.3029,
        "resolution": 0.0132,
        "uncertainty": 0.2197,
        "within_bin_residual": -0.0033
      },
      "plain_language_fr": "Quand le modèle se dit sûr à environ 91 %, il a raison 36 % du temps (sur 606 coups ; plage plausible 33 % à 40 %). En moyenne, il annonce 85 % et réussit 33 % de ses déplacements."
    }
  },
  "games": [
    {
      "seed": 0,
      "scenario_tag": "5327767cb2b8fa45",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 156,
      "hits": 0,
      "bumps": 148,
      "known_wall_repeats": 146,
      "unique_cells": 7,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 156,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 156,
      "brier_score": 0.6615,
      "duration_ms": 87185,
      "evidence_sha256": "ae1c3544c000a43616c05dc51d684d7c52db9ac976680f9d976666dd66c6b1a1"
    },
    {
      "seed": 1,
      "scenario_tag": "123f758b160ea0d4",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 151,
      "hits": 0,
      "bumps": 150,
      "known_wall_repeats": 149,
      "unique_cells": 2,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 151,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 151,
      "brier_score": 0.8009,
      "duration_ms": 99965,
      "evidence_sha256": "a10e7b6fe06c7d1d512d8e5631a6f1d1b96366923283ba0203db48e55b1649ab"
    },
    {
      "seed": 2,
      "scenario_tag": "1054437bd2bcae7b",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 153,
      "hits": 1,
      "bumps": 0,
      "known_wall_repeats": 0,
      "unique_cells": 4,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 153,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 153,
      "brier_score": 0.1667,
      "duration_ms": 109157,
      "evidence_sha256": "61cff7cf4e6d0fc0f7124223ab484bcfd013f3904cb261618c91046746f53b46"
    },
    {
      "seed": 3,
      "scenario_tag": "992847cee040357d",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 210,
      "hits": 0,
      "bumps": 3,
      "known_wall_repeats": 2,
      "unique_cells": 5,
      "revisits": 64,
      "immediate_reversals": 46,
      "llm_calls": 210,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 210,
      "brier_score": 0.0623,
      "duration_ms": 112447,
      "evidence_sha256": "ac503fbdf3f6df8413877074586b588a6612bce51902798b3991f54987459c6b"
    },
    {
      "seed": 4,
      "scenario_tag": "a3775211eaf136e7",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 156,
      "hits": 0,
      "bumps": 149,
      "known_wall_repeats": 147,
      "unique_cells": 7,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 156,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 156,
      "brier_score": 0.6459,
      "duration_ms": 94877,
      "evidence_sha256": "9465fd989a5b6219c552177f902c2137ffe6deba9568d3fbe967532a5cc8437a"
    },
    {
      "seed": 5,
      "scenario_tag": "818402a1c2cdc4ee",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 168,
      "hits": 0,
      "bumps": 2,
      "known_wall_repeats": 1,
      "unique_cells": 5,
      "revisits": 16,
      "immediate_reversals": 10,
      "llm_calls": 168,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 168,
      "brier_score": 0.1191,
      "duration_ms": 86985,
      "evidence_sha256": "84663d5fe8a7c198318ee1fd4fb99613f35a64cde7599a38057b124ee626eef4"
    },
    {
      "seed": 6,
      "scenario_tag": "e3a35b35678367b5",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 203,
      "hits": 0,
      "bumps": 5,
      "known_wall_repeats": 4,
      "unique_cells": 5,
      "revisits": 56,
      "immediate_reversals": 40,
      "llm_calls": 203,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 203,
      "brier_score": 0.1018,
      "duration_ms": 108433,
      "evidence_sha256": "0ff2e57d211db2c31fcd3a176987128df05ee4e5f599fc88a9e55e983802663a"
    },
    {
      "seed": 7,
      "scenario_tag": "f2e49680b9a69523",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 159,
      "hits": 0,
      "bumps": 26,
      "known_wall_repeats": 25,
      "unique_cells": 8,
      "revisits": 2,
      "immediate_reversals": 2,
      "llm_calls": 159,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 159,
      "brier_score": 0.5106,
      "duration_ms": 86467,
      "evidence_sha256": "5094483b20a64f82516a227a864a7d88aa317a2dbd11338e57f03bc8cf332a2c"
    },
    {
      "seed": 8,
      "scenario_tag": "766646d7986ba8ef",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 271,
      "hits": 0,
      "bumps": 12,
      "known_wall_repeats": 11,
      "unique_cells": 3,
      "revisits": 136,
      "immediate_reversals": 126,
      "llm_calls": 271,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 271,
      "brier_score": 0.0564,
      "duration_ms": 151570,
      "evidence_sha256": "ad8e8a1309132459d0432106740d2b742669014b67d9e071c15b4c7d56cc7e23"
    },
    {
      "seed": 9,
      "scenario_tag": "ac93e13596ea6c9a",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 151,
      "hits": 0,
      "bumps": 150,
      "known_wall_repeats": 149,
      "unique_cells": 2,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 151,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 151,
      "brier_score": 0.8009,
      "duration_ms": 101447,
      "evidence_sha256": "0b1846813fb7525e59f3fca137bf8b7e015f64fc4ebb0ef0d125692877154ed9"
    }
  ],
  "report_sha256": "64be3a258afdfdeca0c562093efe78e859fadaefa00c60c0312306db43778c59",
  "timeout_policy": "outilsia.decision_quality.timeout_policy.v2",
  "warmup": {
    "status": "ok",
    "duration_ms": 6934,
    "counted": false
  },
  "honesty_note": "Mesure sur votre machine, graines fixes ; ce n'est pas un classement universel des modèles.",
  "raw_model_output_stored": false
}