{
  "model_details": {
    "provider": "openai",
    "model_requested": "gpt-oss-120b",
    "model_for_requests": "gpt-oss-120b",
    "api_base_url": "",
    "chat_completions_endpoint": ""
  },
  "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": 1006,
    "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": 1006,
    "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": 1006,
    "attempts_with_token_usage": 1006,
    "attempts_with_output_tokens": 1006,
    "attempts_with_cached_input_tokens": 0,
    "attempts_with_thinking_tokens": 0,
    "input_tokens_total": 1216569,
    "cached_input_tokens_total": 0,
    "non_cached_input_tokens_total": 1216569,
    "output_tokens_total": 558185,
    "thinking_tokens_total": 0,
    "output_tokens_definition": "total_tokens - prompt_tokens (or completion_tokens + thinking_tokens fallback)"
  },
  "mode": "metrics_only",
  "source_output_csv": "data/output/like____gptoss120b__2025-11-09-17-09.csv",
  "truth_label_count": 999,
  "prediction_count": 1000,
  "evaluated_example_count": 999,
  "calibration_metrics": {
    "available": true,
    "sample_count": 999,
    "bin_count": 10,
    "ece": 0.25493493493493635,
    "mce": 0.2965565217391315,
    "brier_score": 0.29816296296296374
  },
  "first_prompt_timestamp": "2025-11-09T16:09:03.246413Z",
  "last_prompt_timestamp": "2025-11-09T17:36:51.359054Z",
  "overall_time_seconds": 5268.112641,
  "overall_time_human": "1h 27m 48s",
  "truth_source": "output_csv_truth_column",
  "accuracy": 0.6416416416416416,
  "macro_precision": 0.630263069427628,
  "macro_recall": 0.5384561799320166,
  "macro_f1": 0.5704999879963537,
  "per_label": {
    "Clause-final pragmatic marker": {
      "precision": 0.8860759493670886,
      "recall": 0.5882352941176471,
      "f1": 0.7070707070707072,
      "support": 119
    },
    "Clause-medial pragmatic marker": {
      "precision": 0.7872340425531915,
      "recall": 0.4868421052631579,
      "f1": 0.6016260162601627,
      "support": 380
    },
    "Discourse marker": {
      "precision": 0.6994818652849741,
      "recall": 0.7317073170731707,
      "f1": 0.7152317880794702,
      "support": 369
    },
    "Quotative marker": {
      "precision": 0.7785234899328859,
      "recall": 0.8854961832061069,
      "f1": 0.8285714285714286,
      "support": 131
    },
    "unclassified": {
      "precision": 0,
      "recall": 0,
      "f1": 0,
      "support": 0
    }
  },
  "confusion_matrix": {
    "Clause-final pragmatic marker": {
      "Clause-final pragmatic marker": 70,
      "Clause-medial pragmatic marker": 3,
      "Discourse marker": 38,
      "Quotative marker": 1,
      "unclassified": 7
    },
    "Clause-medial pragmatic marker": {
      "Clause-final pragmatic marker": 4,
      "Clause-medial pragmatic marker": 185,
      "Discourse marker": 69,
      "Quotative marker": 8,
      "unclassified": 114
    },
    "Discourse marker": {
      "Clause-final pragmatic marker": 5,
      "Clause-medial pragmatic marker": 44,
      "Discourse marker": 270,
      "Quotative marker": 24,
      "unclassified": 26
    },
    "Quotative marker": {
      "Clause-final pragmatic marker": 0,
      "Clause-medial pragmatic marker": 3,
      "Discourse marker": 9,
      "Quotative marker": 116,
      "unclassified": 3
    },
    "unclassified": {
      "Clause-final pragmatic marker": 0,
      "Clause-medial pragmatic marker": 0,
      "Discourse marker": 0,
      "Quotative marker": 0,
      "unclassified": 0
    }
  },
  "labels": [
    "Clause-final pragmatic marker",
    "Clause-medial pragmatic marker",
    "Discourse marker",
    "Quotative marker",
    "unclassified"
  ],
  "total_examples": 999,
  "label_metrics_available": true,
  "run_config": {
    "task_name": "*like* discourse/pragm",
    "prompt_layout": null,
    "task_description": "Classification of instance of *like* into four discourse-pragmatic categories: 1. Quotative marker, 2. Discourse marker, 3. Clause-medial pragmatic marker, 4 . Clause-final pragmatic marker",
    "tags": "open source;*like*;discourse;pragmatics;English;v1"
  },
  "source_input_csv": "",
  "source_labels_csv": "",
  "estimated_cost_usd": 1.3797236,
  "pricing": {
    "status": "priced",
    "status_label": "priced",
    "provider_key": "e-infra",
    "matched_key": "gptoss120b",
    "resolved_key": "gpt-oss-120b",
    "pricing_tier": "standard",
    "estimated_cost_usd": 1.3797236,
    "catalog_updated_at": "2026-05-27T15:36:49Z",
    "missing_buckets": []
  }
}