{
  "model_details": {
    "provider": "",
    "model_requested": "gpt-oss-120b",
    "model_for_requests": "gpt-oss-120b",
    "api_base_url": "",
    "chat_completions_endpoint": ""
  },
  "run_config": {
    "task_name": "OE number",
    "prompt_layout": "standard",
    "task_description": "Determining  grammatical number (sg./pl.) in Old English nouns.",
    "tags": "open source;Old English;morphology;number"
  },
  "source_input_csv": "",
  "source_output_csv": "data/output/OE_number____gptoss120b__2026-02-21-18-49.csv",
  "source_labels_csv": "",
  "cache_padding": {
    "enabled": false,
    "target_shared_prefix_tokens": 0,
    "calibration_shared_prefix_tokens": null,
    "target_prompt_tokens": 0,
    "calibration_prompt_tokens": null,
    "calibration_example_id": null,
    "applied_padding_tokens_estimate": 0,
    "examples_with_padding_applied": 0
  },
  "request_control_summary": {
    "configured": {},
    "attempts_total": 1200,
    "attempts_with_control_telemetry": 0,
    "per_control": {
      "reasoning_effort": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "thinking_level": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "effort": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "verbosity": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "prompt_cache_key": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "gemini_cached_content": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "requesty_auto_cache": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      }
    }
  },
  "usage_metadata_summary": {
    "attempts_total": 1200,
    "attempts_with_usage_metadata": 0,
    "attempts_with_cached_token_signals": 0,
    "cached_tokens_total_estimate": 0,
    "cache_read_tokens_total": 0,
    "cache_write_tokens_total": 0,
    "cache_token_fields_totals": {},
    "attempts_with_gemini_cached_content_token_signals": 0,
    "gemini_cached_content_token_count_total": 0,
    "gemini_cached_content_token_fields_totals": {}
  },
  "token_usage_totals": {
    "attempts_total": 1200,
    "attempts_with_token_usage": 1200,
    "attempts_with_output_tokens": 1200,
    "attempts_with_cached_input_tokens": 0,
    "attempts_with_thinking_tokens": 0,
    "input_tokens_total": 923408,
    "cached_input_tokens_total": 0,
    "non_cached_input_tokens_total": 923408,
    "output_tokens_total": 555000,
    "thinking_tokens_total": 0,
    "output_tokens_definition": "total_tokens - prompt_tokens (or completion_tokens + thinking_tokens fallback)"
  },
  "mode": "metrics_only",
  "truth_label_count": 1200,
  "prediction_count": 1200,
  "evaluated_example_count": 1200,
  "calibration_metrics": {
    "available": true,
    "sample_count": 1200,
    "bin_count": 10,
    "ece": 0.06971666666667636,
    "mce": 0.4,
    "brier_score": 0.10843149999999999
  },
  "first_prompt_timestamp": "2026-02-21T17:49:34.964163Z",
  "last_prompt_timestamp": "2026-02-21T21:20:52.156285Z",
  "overall_time_seconds": 12677.192122,
  "overall_time_human": "3h 31m 17s",
  "truth_source": "output_csv_truth_column",
  "accuracy": 0.8766666666666667,
  "macro_precision": 0.6029494830423144,
  "macro_recall": 0.5902634358594377,
  "macro_f1": 0.5953509904045788,
  "per_label": {
    "plural": {
      "precision": 0.8401360544217688,
      "recall": 0.9014598540145985,
      "f1": 0.869718309859155,
      "support": 274
    },
    "singular": {
      "precision": 0.9687123947051745,
      "recall": 0.8693304535637149,
      "f1": 0.9163346613545815,
      "support": 926
    },
    "unclassified": {
      "precision": 0,
      "recall": 0,
      "f1": 0,
      "support": 0
    }
  },
  "labels": [
    "plural",
    "singular",
    "unclassified"
  ],
  "label_count": 3,
  "total_examples": 1200,
  "confusion_matrix": {
    "plural": {
      "plural": 247,
      "singular": 26,
      "unclassified": 1
    },
    "singular": {
      "plural": 47,
      "singular": 805,
      "unclassified": 74
    },
    "unclassified": {
      "plural": 0,
      "singular": 0,
      "unclassified": 0
    }
  },
  "label_metrics_available": true,
  "estimated_cost_usd": 1.2573632,
  "pricing": {
    "status": "priced",
    "status_label": "priced",
    "provider_key": "e-infra",
    "matched_key": "gptoss120b",
    "resolved_key": "gpt-oss-120b",
    "pricing_tier": "standard",
    "estimated_cost_usd": 1.2573632,
    "catalog_updated_at": "2026-05-27T15:36:49Z",
    "missing_buckets": []
  }
}