{
  "model_details": {
    "provider": "vertex",
    "model_requested": "gemini-3.1-flash-lite-preview",
    "model_for_requests": "google/gemini-3.1-flash-lite-preview",
    "api_base_url": "https://aiplatform.googleapis.com/v1/projects/gen-lang-client-0461875984/locations/global/endpoints/openapi",
    "chat_completions_endpoint": "https://aiplatform.googleapis.com/v1/projects/gen-lang-client-0461875984/locations/global/endpoints/openapi/chat/completions",
    "vertex_cache_model": "projects/gen-lang-client-0461875984/locations/global/publishers/google/models/gemini-3.1-flash-lite-preview"
  },
  "run_config": {
    "input": [
      "data/input/sentiment.csv"
    ],
    "labels": "",
    "task_name": "sentiment analysis",
    "task_description": "Sentiment analysis of the Movie Review Data (https://www.cs.cornell.edu/people/pabo/movie-review-data/). The movie review excerpts are labelled as positive or negative.",
    "tags": "sentiment analysis; English; semantics; pragmatics",
    "model": "gemini-3.1-flash-lite-preview",
    "temperature": null,
    "top_p": null,
    "top_k": null,
    "service_tier": "standard",
    "verbosity": null,
    "reasoning_effort": null,
    "thinking_level": null,
    "effort": null,
    "strict_control_acceptance": true,
    "provider": "vertex",
    "system_prompt": "You will analyze an excerpt representing an online movie review and decide whether the excerpt represents a positive or a negative sentiment of the reviewer towards the movie.\nIn case of a positive sentiment, return \"positive\", in case of a negative sentiment, return \"negative\".",
    "system_prompt_b64": null,
    "few_shot_examples": 0,
    "prompt_layout": "standard",
    "cache_pad_target_tokens": 0,
    "prompt_cache_key": null,
    "gemini_cached_content": null,
    "requesty_auto_cache": null,
    "vertex_auto_adc_login": null,
    "vertex_access_token_refresh_seconds": null,
    "create_gemini_cache": false,
    "gemini_cache_ttl": 3600,
    "gemini_cache_ttl_autoupdate": true,
    "keep_gemini_cache": false,
    "enable_cot": true,
    "no_explanation": false,
    "logprobs": false,
    "calibration": true,
    "confusion_heatmap": true,
    "api_key_var": "VERTEX_ACCESS_TOKEN",
    "api_base_var": "VERTEX_BASE_URL",
    "max_retries": 3,
    "retry_delay": 5.0,
    "request_interval_ms": 0,
    "request_timeout_seconds": 600.0,
    "threads": 4,
    "prompt_log_detail": "full",
    "flush_rows": 100,
    "flush_seconds": 2.0,
    "validator_cmd": null,
    "validator_args": "",
    "validator_timeout": 5.0,
    "validator_prompt_max_candidates": 50,
    "validator_prompt_max_chars": 8000,
    "validator_exhausted_policy": "accept_blank_confidence",
    "validator_debug": false,
    "log_level": "INFO"
  },
  "source_input_csv": "data/input/sentiment.csv",
  "source_output_csv": "data/output/sentiment__vertex__gemini31flashlitepreview__2026-04-29-23-18.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": 417,
    "attempts_with_control_telemetry": 417,
    "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": 417,
    "attempts_with_usage_metadata": 417,
    "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": 417,
    "attempts_with_token_usage": 417,
    "attempts_with_output_tokens": 417,
    "attempts_with_cached_input_tokens": 0,
    "attempts_with_thinking_tokens": 0,
    "input_tokens_total": 231723,
    "cached_input_tokens_total": 0,
    "non_cached_input_tokens_total": 231723,
    "output_tokens_total": 51353,
    "thinking_tokens_total": 0,
    "output_tokens_definition": "total_tokens - prompt_tokens (or completion_tokens + thinking_tokens fallback)"
  },
  "mode": "metrics_only",
  "truth_label_count": 400,
  "prediction_count": 400,
  "evaluated_example_count": 400,
  "calibration_metrics": {
    "available": true,
    "sample_count": 392,
    "bin_count": 10,
    "ece": 0.02451530612245119,
    "mce": 0.75,
    "brier_score": 0.044052806122448974
  },
  "first_prompt_timestamp": "2026-04-29T21:18:08.340230Z",
  "last_prompt_timestamp": "2026-04-29T21:20:57.771719Z",
  "overall_time_seconds": 169.431489,
  "overall_time_human": "2m 49s",
  "truth_source": "output_csv_truth_column",
  "accuracy": 0.9475,
  "cohen_kappa": 0.8952618453865336,
  "macro_precision": 0.6335412579438939,
  "macro_recall": 0.6316666666666667,
  "macro_f1": 0.6324346005841136,
  "per_label": {
    "negative": {
      "precision": 0.9368932038834952,
      "recall": 0.965,
      "f1": 0.9507389162561576,
      "support": 200
    },
    "positive": {
      "precision": 0.9637305699481865,
      "recall": 0.93,
      "f1": 0.9465648854961832,
      "support": 200
    },
    "unclassified": {
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "support": 0
    }
  },
  "labels": [
    "negative",
    "positive",
    "unclassified"
  ],
  "label_count": 3,
  "total_examples": 400,
  "confusion_matrix": {
    "negative": {
      "negative": 193,
      "positive": 7,
      "unclassified": 0
    },
    "positive": {
      "negative": 13,
      "positive": 186,
      "unclassified": 1
    },
    "unclassified": {
      "negative": 0,
      "positive": 0,
      "unclassified": 0
    }
  },
  "label_metrics_available": true
}