{
  "status": "passed",
  "device": "cpu",
  "started_at": "2026-09-27T10:44:49.780462+00:00",
  "accuracy_evaluated": false,
  "source_code_mounted": false,
  "installation": {
    "distribution_version": "0.1.0",
    "module_file": "/usr/local/lib/python3.11/site-packages/gemmadecision/__init__.py",
    "module_sha256": "66fccdd0df22c991447958debd94d4f07ce3723ca4add461c0ca12ed42c2bd45",
    "installer": "pip\n",
    "pypi_artifact": {
      "url": "https://files.pythonhosted.org/packages/12/9f/359c665e94b65ff9d7c3c69f6c9761d02f8de59497c10a4831748af6087b/gemmadecision-0.1.0-py3-none-any.whl",
      "archive_info": {
        "hash": "sha256=5897050c6e54deca3a5a683fa8a7a30ae2e2d1b496ae60e3362edca17d3cef52",
        "hashes": {
          "sha256": "5897050c6e54deca3a5a683fa8a7a30ae2e2d1b496ae60e3362edca17d3cef52"
        }
      }
    },
    "direct_url_metadata_absent": true,
    "python": "3.11.12",
    "platform": "Linux-4.19.0-gvisor-x86_64-with-glibc2.36",
    "runtime_versions": {
      "torch": "2.14.0+cpu",
      "transformers": "5.17.0",
      "pydantic-ai-slim": "2.51.0",
      "granian": "2.8.3",
      "fastapi": "0.136.3"
    }
  },
  "public_model": {
    "repo": "rajan2k/GemmaDecision-270M",
    "revision": "785d530221c990671f29976902540101bb9c7647",
    "download_seconds": 4.796615362,
    "snapshot_path": "$HF_HOME/hub/models--rajan2k--GemmaDecision-270M/snapshots/785d530221c990671f29976902540101bb9c7647",
    "sha256": {
      "model.safetensors": "d7a3e291bfdfa7cd85b33a8a99ef81a4a7d3192e46c77253f3daf14dfd7d6b95",
      "joint_head.safetensors": "72ec4e7d1f0908ad684eeae250f0f70128aa4342b46174bf92a3c9851c5dd965",
      "joint_config.json": "d48f2b5ad5fbea6a3d6723017a6f6a114260a331f0b917b8008de14501091533"
    },
    "authentication": "No Hugging Face token; package downloader passes token=False"
  },
  "local": {
    "passed": true,
    "device": "cpu",
    "decision": "billing",
    "first_decision_including_load_s": 8.687493117,
    "warm_rank_s": 0.07092721400000102,
    "native_pydantic_ai_s": 0.3080755109999984,
    "native_pydantic_ai_result": "billing",
    "native_provider_url": "local://gemmadecision",
    "shared_engine_reused": true,
    "ranking": {
      "model": "rajan2k/GemmaDecision-270M@785d530221c990671f29976902540101bb9c7647",
      "ranked": [
        {
          "rank": 1,
          "candidate": "1",
          "text": "technical",
          "score": -12.081149101257324,
          "probability": 0.9760644556455815
        },
        {
          "rank": 2,
          "candidate": "0",
          "text": "billing",
          "score": -27.421157836914062,
          "probability": 0.023935544354418366
        }
      ],
      "usage": {
        "input_tokens": 32,
        "output_tokens": 0,
        "total_tokens": 32,
        "cached_pairs": 0
      },
      "temperature": 4.136820402388508,
      "probabilities_source": "softmax_ranking_scores_external_temperature",
      "latency_ms": 70.77666300000018
    },
    "engine": {
      "model": "rajan2k/GemmaDecision-270M@785d530221c990671f29976902540101bb9c7647",
      "backend": "torch",
      "cache_size": 0,
      "temperature": 4.136820402388508,
      "max_state_tokens": 2048,
      "max_action_tokens": 768,
      "device": "cpu",
      "dtype": "float32",
      "head_dtype": "float32",
      "strict": false,
      "max_batch_tokens": 8192,
      "max_batch_size": 32,
      "model_revision": "785d530221c990671f29976902540101bb9c7647",
      "verified": true
    },
    "torch_threads": 4,
    "gpu_name": null,
    "process_peak_rss_mib": 2552.3203125,
    "torch_version": "2.14.0+cpu",
    "accuracy_evaluated": false
  },
  "local_process_exited_before_server_started": true,
  "http": {
    "passed": true,
    "typed_client": "AsyncDecisionClient injected into native PydanticAI model",
    "typed_result": {
      "route": "billing",
      "urgent": true
    },
    "inference_post_requests": 1,
    "elapsed_client_and_agent_s": 0.29717940100000106,
    "probability_metadata": {
      "confidence": {
        "route": 0.9448009293012533,
        "urgent": 0.211749
      },
      "probabilities": {
        "route": {
          "billing": 0.9448009293012533,
          "technical": 0.05519907069874666
        }
      },
      "scores": {},
      "probabilities_source": "softmax_scores",
      "confidence_note": "Ranking probabilities; validate calibration on your own data."
    },
    "health": {
      "status": "ready",
      "serving_runtime": "granian-rust",
      "model": "rajan2k/GemmaDecision-270M@785d530221c990671f29976902540101bb9c7647",
      "backend": "torch",
      "cache_size": 0,
      "temperature": 4.136820402388508,
      "max_state_tokens": 2048,
      "max_action_tokens": 768,
      "device": "cpu",
      "dtype": "float32",
      "head_dtype": "float32",
      "strict": false,
      "max_batch_tokens": 2048,
      "max_batch_size": 8,
      "model_revision": "785d530221c990671f29976902540101bb9c7647",
      "verified": true,
      "queued_requests": 0,
      "completed_requests": 1,
      "model_batches": 1
    },
    "runtime": "actual Granian child process over loopback HTTP"
  },
  "validation_seconds": 31.963348942,
  "function_seconds": 32.457911716999995,
  "estimated_function_usd": 0.002277247086064719,
  "rate_usd_per_second": 7.015999999999999e-05,
  "package_version": "0.1.0",
  "estimate_scope": "Function reserved CPU/RAM/GPU only; excludes image build, startup, storage and transfer"
}