{
  "artifacts": [
    {
      "file": "08ac69d8aab391c36d1918c3bbce48ce3aea646f4a88457552904482952db3c7.tflite",
      "sha256": "628eb9b3b8be096b57f626c7a0e01403781c3dd37466bea7bcdaf3e63076e45d",
      "size_mb": 8.41
    }
  ],
  "benchmarks": [],
  "browser": {
    "backends": [
      {
        "backend": "wasm_xnnpack",
        "date": "2026-08-11",
        "env": {
          "browser": "chromium",
          "browser_version": "151.0.7922.34",
          "headless": true,
          "jspi": true,
          "litertjs_core_version": "2.5.3",
          "machine_label": "mac-studio-m4-max",
          "os": "macOS",
          "os_version": "27.0.0",
          "webgpu_adapter": {
            "architecture": "metal-3",
            "description": "",
            "device": "",
            "vendor": "apple"
          }
        },
        "full_delegation": null,
        "latency_p50_ms": 432.95,
        "loads": true,
        "max_rel_diff": null,
        "output_match": null,
        "provenance": "measured",
        "runs": true
      },
      {
        "backend": "webgpu_mldrift",
        "date": "2026-08-11",
        "env": {
          "browser": "chromium",
          "browser_version": "151.0.7922.34",
          "headless": true,
          "jspi": true,
          "litertjs_core_version": "2.5.3",
          "machine_label": "mac-studio-m4-max",
          "os": "macOS",
          "os_version": "27.0.0",
          "webgpu_adapter": {
            "architecture": "metal-3",
            "description": "",
            "device": "",
            "vendor": "apple"
          }
        },
        "full_delegation": true,
        "latency_p50_ms": 25.665,
        "loads": true,
        "max_rel_diff": 43274.63062780912,
        "output_match": false,
        "provenance": "measured",
        "runs": true
      }
    ],
    "demo_url": null,
    "sweep_source": "data/sweep/2.5.3/2026-08-11/d-fine-s__dfine_graphb_fp16.json"
  },
  "conversion": {
    "command": "python scripts/build_dfine_split.py && python scripts/build_dfine_fix3.py && python scripts/pack_assets.py",
    "quantization": "fp16",
    "tool": "litert-torch",
    "tool_version": "0.10.0 (editable dev checkout 115a136 + local patches)"
  },
  "cross_runtime": [],
  "delegation": null,
  "device": {
    "records": [
      {
        "device": "galaxy-s26",
        "run": {
          "accelerator": "gpu_mldrift",
          "context_length": null,
          "date": "2026-08-25",
          "decode_tokens_per_s": null,
          "delegated_ops": null,
          "env": {
            "device": "Galaxy S26 (SM-S942Q)",
            "machine_label": "galaxy-s26-npubench",
            "os_build": "Android 16",
            "runtime": "litert",
            "runtime_version": "2.2.0",
            "soc": "Qualcomm SM8850",
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "npubench sweep row: N=50 median, 1 backend = 1 process, accepted only with thermal NONE->NONE; NPU rows additionally required qnn_partition delegate evidence in logcat (a stock model asked for on the NPU can silently land on XNNPACK and report a CPU number)",
            "mode=jit thermal=NONE->NONE headroom=0.78000003->0.79666674 attempt=0"
          ],
          "failure_class": null,
          "full_delegation": null,
          "latency_p50_ms": 134.036,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "iterations": 50,
            "latency_max_ms": 135.795,
            "latency_min_ms": 132.853,
            "load_ms": 1531.1
          },
          "output_match": null,
          "peak_mem_mb": null,
          "prefill_tokens_per_s": null,
          "provenance": "measured",
          "runs": true,
          "total_ops": null,
          "ttft_ms": null
        },
        "source": "data/device_runs/2.2.0/2026-08-25/d-fine-s__dfine_graphb_fp16__galaxy-s26.json"
      },
      {
        "device": "galaxy-s26",
        "run": {
          "accelerator": "npu_qnn",
          "context_length": null,
          "date": "2026-08-25",
          "decode_tokens_per_s": null,
          "delegated_ops": null,
          "env": {
            "device": "Galaxy S26 (SM-S942Q)",
            "machine_label": "galaxy-s26-npubench",
            "os_build": "Android 16",
            "runtime": "litert",
            "runtime_version": "2.2.0",
            "soc": "Qualcomm SM8850",
            "vendor_sdk": "QAIRT (Hexagon, JIT on-device)"
          },
          "error": null,
          "evidence": [
            "npubench sweep row: N=50 median, 1 backend = 1 process, accepted only with thermal NONE->NONE; NPU rows additionally required qnn_partition delegate evidence in logcat (a stock model asked for on the NPU can silently land on XNNPACK and report a CPU number)",
            "mode=jit thermal=NONE->NONE headroom=0.7633334->0.78000003 attempt=0"
          ],
          "failure_class": null,
          "full_delegation": null,
          "latency_p50_ms": 172.363,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "iterations": 50,
            "latency_max_ms": 173.358,
            "latency_min_ms": 169.783,
            "load_ms": 184.2,
            "qnn_partition_lines": 57
          },
          "output_match": null,
          "peak_mem_mb": null,
          "prefill_tokens_per_s": null,
          "provenance": "measured",
          "runs": true,
          "total_ops": null,
          "ttft_ms": null
        },
        "source": "data/device_runs/2.2.0/2026-08-25/d-fine-s__dfine_graphb_fp16__galaxy-s26.json"
      },
      {
        "device": "raspberry-pi-5",
        "run": {
          "accelerator": "cpu_xnnpack",
          "context_length": null,
          "date": "2026-08-31",
          "decode_tokens_per_s": null,
          "delegated_ops": null,
          "env": {
            "device": "Raspberry Pi 5 Model B Rev 1.1",
            "machine_label": "raspberry-pi-5",
            "os_build": "Linux-6.18.34+rpt-rpi-2712-aarch64-with-glibc2.41",
            "runtime": "litert",
            "runtime_version": "2.2.0.dev20260804",
            "soc": null,
            "vendor_sdk": null
          },
          "error": null,
          "evidence": [
            "pi5 sweep row: LiteRT benchmark_model, CPU/XNNPACK at --num_threads=4, 3 invocations per file of 10 warm-up + 50 timed runs (the tool caps a phase at 150 s, so very slow graphs run fewer); latency = median of the three per-invocation medians over the timed phase; a row counts as measured only when every invocation exited 0 with XNNPACK engaged and vcgencmd get_throttled 0x0 before and after",
            "versions: ai-edge-litert-nightly=2.2.0.dev20260804, benchmark_model_sha256=babc9275addd8612caf0842294cb5b3f301c3cce46e6c5979cd2803f215f6bee, cpu=Raspberry Pi 5 Model B Rev 1.1, litert-cli-nightly=0.2.0.dev20260805, platform=Linux-6.18.34+rpt-rpi-2712-aarch64-with-glibc2.41, python=3.13.5",
            "invocation 0: exit=0 wall_s=61.3 temp 51.0->59.8C throttled=0x0 xnnpack=True median_us=1018760.0 runs=50 footprint_peak_mb=426.64",
            "invocation 1: exit=0 wall_s=62.2 temp 51.6->58.2C throttled=0x0 xnnpack=True median_us=1034528.0 runs=50 footprint_peak_mb=426.59",
            "invocation 2: exit=0 wall_s=60.9 temp 50.5->58.7C throttled=0x0 xnnpack=True median_us=1013701.0 runs=50 footprint_peak_mb=426.59"
          ],
          "failure_class": null,
          "full_delegation": null,
          "latency_p50_ms": 1018.76,
          "loads": true,
          "max_abs_diff": null,
          "max_rel_diff": null,
          "metrics": {
            "init_ms": 25.332,
            "invocations": 3,
            "iterations": 150,
            "latency_max_ms": 1038.593,
            "latency_min_ms": 1002.508,
            "threads": 4,
            "warmup_runs_per_invocation": 10
          },
          "output_match": null,
          "peak_mem_mb": 426.64,
          "prefill_tokens_per_s": null,
          "provenance": "measured",
          "runs": true,
          "total_ops": null,
          "ttft_ms": null
        },
        "source": "data/device_runs/2.2.0.dev20260804/2026-08-31/d-fine-s__dfine_graphb_fp16__raspberry-pi-5.json"
      }
    ]
  },
  "model": {
    "family": "d-fine",
    "id": "d-fine-s__dfine_graphb_fp16",
    "license": "apache-2.0",
    "source_url": "https://huggingface.co/litert-community/D-FINE-S-LiteRT",
    "task": "object-detection"
  },
  "pitfalls": [
    "Graph B consumes the host-computed top-300 selection (memory, target, ref); sigmoid + threshold + cxcywh->xyxy + light NMS are host-side (HF card 'How it runs')."
  ],
  "schema_version": "1.2"
}