{
  "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": "qwen3:0.6b",
    "quantization": "Q4_K_M",
    "parameter_size": "751.63M",
    "family": "qwen3",
    "digest_prefix": "7df6b6e09427"
  },
  "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": 1791548831525,
  "finished_at_ms": 1791549552170,
  "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": 150.0,
    "mean_hits": 0.0,
    "mean_bumps": 21.8,
    "known_wall_repeats": 208,
    "outcomes": {
      "timeout": 10
    },
    "llm_calls": 1500,
    "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": 478.0,
    "calibration": {
      "definition": "P(le déplacement annoncé ne heurte pas de mur) ; déplacements hors pass",
      "samples": 218,
      "brier_score": 0.7747,
      "brier_reference_constant_half": 0.25,
      "mean_confidence": 0.8392,
      "observed_move_success_rate": 0.0,
      "overconfidence": 0.8392,
      "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": 0.0,
      "observed_rate_ci95_high": 0.0173,
      "reliability_bins": [
        {
          "lower": 0.0,
          "upper": 0.1,
          "samples": 17,
          "mean_confidence": 0.0,
          "observed_rate": 0.0,
          "observed_rate_ci95_low": 0.0,
          "observed_rate_ci95_high": 0.1843,
          "gap": 0.0,
          "sufficient": true
        },
        {
          "lower": 0.5,
          "upper": 0.6,
          "samples": 11,
          "mean_confidence": 0.5,
          "observed_rate": 0.0,
          "observed_rate_ci95_low": 0.0,
          "observed_rate_ci95_high": 0.2588,
          "gap": 0.5,
          "sufficient": true
        },
        {
          "lower": 0.9,
          "upper": 1.0,
          "samples": 190,
          "mean_confidence": 0.9339,
          "observed_rate": 0.0,
          "observed_rate_ci95_low": 0.0,
          "observed_rate_ci95_high": 0.0198,
          "gap": 0.9339,
          "sufficient": true
        }
      ],
      "ece": 0.8392,
      "ece_ci95_low": 0.8036,
      "ece_ci95_high": 0.8738,
      "mce": 0.9339,
      "murphy": {
        "reliability": 0.7728,
        "resolution": 0.0,
        "uncertainty": 0.0,
        "within_bin_residual": 0.0019
      },
      "plain_language_fr": "Quand le modèle se dit sûr à environ 93 %, il a raison 0 % du temps (sur 190 coups ; plage plausible 0 % à 2 %). En moyenne, il annonce 84 % et réussit 0 % de ses déplacements."
    }
  },
  "games": [
    {
      "seed": 0,
      "scenario_tag": "5327767cb2b8fa45",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 32,
      "known_wall_repeats": 31,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.7554,
      "duration_ms": 71457,
      "evidence_sha256": "367cd9120b9832c0a00b22be2a6bbc17ca85f906aebd3c51dfea7544f6f43013"
    },
    {
      "seed": 1,
      "scenario_tag": "123f758b160ea0d4",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 24,
      "known_wall_repeats": 23,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.7924,
      "duration_ms": 71254,
      "evidence_sha256": "709909bee892cd953de8b172cf81261d39adb30c52d6852d38b00d64c20510ea"
    },
    {
      "seed": 2,
      "scenario_tag": "1054437bd2bcae7b",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 13,
      "known_wall_repeats": 12,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.7968,
      "duration_ms": 71260,
      "evidence_sha256": "6fdb1ca59d11d65d447ad84ff95e8f273ec3522ddab2d55a9a73a6df54da1b40"
    },
    {
      "seed": 3,
      "scenario_tag": "992847cee040357d",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 33,
      "known_wall_repeats": 32,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.7981,
      "duration_ms": 70889,
      "evidence_sha256": "b782cb86c42d3434c5905ef2f50acaddb8f3041c45688de1cf7064892b6b49b6"
    },
    {
      "seed": 4,
      "scenario_tag": "a3775211eaf136e7",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 14,
      "known_wall_repeats": 13,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.6814,
      "duration_ms": 71178,
      "evidence_sha256": "7506c2d3167dc4666882ccc4f71f63db2f967f30d5a69d7bf2d355df50909df3"
    },
    {
      "seed": 5,
      "scenario_tag": "818402a1c2cdc4ee",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 12,
      "known_wall_repeats": 11,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.6965,
      "duration_ms": 72922,
      "evidence_sha256": "c76143e243165f24de8fac9b703d94aea3a344b9e400c77fe165388ae6d4fa1e"
    },
    {
      "seed": 6,
      "scenario_tag": "e3a35b35678367b5",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 21,
      "known_wall_repeats": 20,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.7424,
      "duration_ms": 72320,
      "evidence_sha256": "4557923ba7b35f0ca3f616d1359cad3c94a0bdf876fa91a1362955fc7d71dd5c"
    },
    {
      "seed": 7,
      "scenario_tag": "f2e49680b9a69523",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 16,
      "known_wall_repeats": 15,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.8306,
      "duration_ms": 72836,
      "evidence_sha256": "f88f71a83bd28665d483d00f91bf260ea9326e5242b3b800dac0f9c635035f63"
    },
    {
      "seed": 8,
      "scenario_tag": "766646d7986ba8ef",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 24,
      "known_wall_repeats": 23,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.7741,
      "duration_ms": 72362,
      "evidence_sha256": "540c290097b1e6bcc5e10857c0f12cee9addf28c1782591584d92ea9da97caf7"
    },
    {
      "seed": 9,
      "scenario_tag": "ac93e13596ea6c9a",
      "outcome": "timeout",
      "won": false,
      "picked_treasure": false,
      "rounds": 150,
      "decisions": 150,
      "hits": 0,
      "bumps": 29,
      "known_wall_repeats": 28,
      "unique_cells": 1,
      "revisits": 0,
      "immediate_reversals": 0,
      "llm_calls": 150,
      "invalid_responses": 0,
      "timed_out_calls": 0,
      "timeout_retries": 0,
      "timeouts_after_retry": 0,
      "forced_passes": 0,
      "dropped_probability_fields": 0,
      "confidence_samples": 150,
      "brier_score": 0.8154,
      "duration_ms": 71049,
      "evidence_sha256": "a7e2126eb6a2c00a9fa1807b15db8a020350b159edcee7759fe71dd66d61582c"
    }
  ],
  "report_sha256": "dc251b237a6b96530e4d70d01aee8ccbaa55e533d8f9c16af8f402282dbfc6f4",
  "timeout_policy": "outilsia.decision_quality.timeout_policy.v2",
  "warmup": {
    "status": "ok",
    "duration_ms": 2920,
    "counted": false
  },
  "honesty_note": "Mesure sur votre machine, graines fixes ; ce n'est pas un classement universel des modèles.",
  "raw_model_output_stored": false
}