{
  "schema_version": 1,
  "config": {
    "in_ch": 3,
    "cond_ch": 7,
    "fine": 128,
    "base": 64,
    "tiny": false,
    "params": 7565251,
    "steps": 20,
    "batches": [
      1,
      8,
      32
    ],
    "parity_batches": [
      1,
      4,
      8
    ],
    "warmup": 10,
    "iters": 100,
    "sample_iters": 30,
    "seed": 0,
    "tf32": "default",
    "parity_only": false,
    "allow_tf32_matmul": false,
    "allow_tf32_cudnn": true
  },
  "weights": {
    "weights": "random",
    "ckpt_sha256": null
  },
  "env": {
    "torch": "2.14.0+cu130",
    "python": "3.11.15",
    "device": "cuda",
    "gpu": "NVIDIA A100-SXM4-40GB",
    "onnxruntime": "1.30.0",
    "git_sha": "ecaba28"
  },
  "backends": {
    "eager": {
      "status": "ok",
      "note": "fp32, torch.no_grad",
      "parity_max_abs_err": {
        "1": 0.0,
        "4": 0.0,
        "8": 0.0
      },
      "velocity": {
        "1": {
          "p50_ms": 8.078434504568577,
          "p99_ms": 13.323198030702775,
          "mean_ms": 8.185998359695077,
          "iters": 100,
          "samples_per_s": 122.1598094770751
        },
        "8": {
          "p50_ms": 11.133172432892025,
          "p99_ms": 11.233665719628334,
          "mean_ms": 11.13595889415592,
          "iters": 100,
          "samples_per_s": 718.3934563729712
        },
        "32": {
          "p50_ms": 35.50215100403875,
          "p99_ms": 35.594888960476965,
          "mean_ms": 35.506853356491774,
          "iters": 100,
          "samples_per_s": 901.2344653218727
        }
      },
      "sample": {
        "1": {
          "p50_ms": 91.69737598858774,
          "p99_ms": 159.82650767546147,
          "mean_ms": 99.47222737440218,
          "iters": 30,
          "samples_per_s": 10.05305728438264
        },
        "8": {
          "p50_ms": 218.73654692899436,
          "p99_ms": 218.85726878419518,
          "mean_ms": 218.7398669697965,
          "iters": 30,
          "samples_per_s": 36.573122727118765
        },
        "32": {
          "p50_ms": 706.5022919559851,
          "p99_ms": 707.3480078321882,
          "mean_ms": 706.5973530523479,
          "iters": 30,
          "samples_per_s": 45.28746090226196
        }
      },
      "post_projection": {
        "nvr": 0.0,
        "mce_max": 7.450580596923828e-08
      }
    },
    "compile": {
      "status": "ok",
      "note": "torch.compile default mode",
      "parity_max_abs_err": {
        "1": 0.0013611316680908203,
        "4": 0.0010221600532531738,
        "8": 0.0009836554527282715
      },
      "velocity": {
        "1": {
          "p50_ms": 2.737137023359537,
          "p99_ms": 2.7649707556702197,
          "mean_ms": 2.738589846994728,
          "iters": 100,
          "samples_per_s": 365.1514304332134
        },
        "8": {
          "p50_ms": 7.656070403754711,
          "p99_ms": 7.676070088054985,
          "mean_ms": 7.657413277775049,
          "iters": 100,
          "samples_per_s": 1044.739223259541
        },
        "32": {
          "p50_ms": 26.16294741164893,
          "p99_ms": 26.19922718498856,
          "mean_ms": 26.1632509669289,
          "iters": 100,
          "samples_per_s": 1223.0895938906415
        }
      },
      "sample": {
        "1": {
          "p50_ms": 49.06173900235444,
          "p99_ms": 49.69799069222063,
          "mean_ms": 48.917095782235265,
          "iters": 30,
          "samples_per_s": 20.44275082175177
        },
        "8": {
          "p50_ms": 148.40310299769044,
          "p99_ms": 148.79322552122176,
          "mean_ms": 148.4296794515103,
          "iters": 30,
          "samples_per_s": 53.897576479059076
        },
        "32": {
          "p50_ms": 519.074086099863,
          "p99_ms": 519.1778369573876,
          "mean_ms": 519.0730291496342,
          "iters": 30,
          "samples_per_s": 61.64835813647197
        }
      },
      "post_projection": {
        "nvr": 0.0,
        "mce_max": 7.450580596923828e-08
      }
    },
    "aoti": {
      "status": "ok",
      "note": "torch.export, dynamic batch 1..4096, aoti_compile_and_package",
      "parity_max_abs_err": {
        "1": 0.0009888708591461182,
        "4": 0.0010471940040588379,
        "8": 0.0010083764791488647
      },
      "velocity": {
        "1": {
          "p50_ms": 2.2335489047691226,
          "p99_ms": 2.2537241200916474,
          "mean_ms": 2.2350755683146417,
          "iters": 100,
          "samples_per_s": 447.41216546608746
        },
        "8": {
          "p50_ms": 7.459083921276033,
          "p99_ms": 7.495929335709662,
          "mean_ms": 7.464080147910863,
          "iters": 100,
          "samples_per_s": 1071.799852288448
        },
        "32": {
          "p50_ms": 25.95880243461579,
          "p99_ms": 25.981307656038553,
          "mean_ms": 25.9605278307572,
          "iters": 100,
          "samples_per_s": 1232.6405768255386
        }
      },
      "sample": {
        "1": {
          "p50_ms": 43.317668489180505,
          "p99_ms": 44.231014982797205,
          "mean_ms": 43.517972606544696,
          "iters": 30,
          "samples_per_s": 22.979011661255775
        },
        "8": {
          "p50_ms": 148.42038799542934,
          "p99_ms": 148.95002960925922,
          "mean_ms": 148.47066898364574,
          "iters": 30,
          "samples_per_s": 53.88269652695652
        },
        "32": {
          "p50_ms": 518.4656314086169,
          "p99_ms": 518.9790421584621,
          "mean_ms": 518.5500039874265,
          "iters": 30,
          "samples_per_s": 61.71053852846159
        }
      },
      "post_projection": {
        "nvr": 0.0,
        "mce_max": 0.0
      }
    },
    "ort": {
      "status": "ok",
      "note": "onnxruntime CUDAExecutionProvider, legacy exporter, opset 18, host copies included",
      "parity_max_abs_err": {
        "1": 0.0011383891105651855,
        "4": 0.0015149116516113281,
        "8": 0.0015028715133666992
      },
      "velocity": {
        "1": {
          "p50_ms": 3.37083893828094,
          "p99_ms": 3.545884496998042,
          "mean_ms": 3.381021940149367,
          "iters": 100,
          "samples_per_s": 295.7685627901669
        },
        "8": {
          "p50_ms": 12.355459504760802,
          "p99_ms": 12.81292530009523,
          "mean_ms": 12.38884152378887,
          "iters": 100,
          "samples_per_s": 645.742379110954
        },
        "32": {
          "p50_ms": 42.81621042173356,
          "p99_ms": 43.282888589892536,
          "mean_ms": 42.833039516117424,
          "iters": 100,
          "samples_per_s": 747.0868367387022
        }
      },
      "sample": {
        "1": {
          "p50_ms": 65.47883455641568,
          "p99_ms": 66.7918698838912,
          "mean_ms": 65.55331978015602,
          "iters": 30,
          "samples_per_s": 15.254757552381276
        },
        "8": {
          "p50_ms": 248.12883604317904,
          "p99_ms": 254.72736506722867,
          "mean_ms": 248.5872142094498,
          "iters": 30,
          "samples_per_s": 32.18186432251304
        },
        "32": {
          "p50_ms": 857.3480820050463,
          "p99_ms": 859.5663432846777,
          "mean_ms": 857.5706561872115,
          "iters": 30,
          "samples_per_s": 37.3147096033732
        }
      },
      "post_projection": {
        "nvr": 0.0,
        "mce_max": 7.450580596923828e-08
      }
    }
  }
}