# Стенд-проверка (stage-2) · промт оператора:

(универсальный)

Ответ агента:

{
  "worked": true,
  "confidence": 7,
  "verdict": "Built matched direct particle-query and exact-moment particle-field neural systems on a structurally matched custom point-cloud regression track. Across 8 paired seeds, the moment system won on every seed: mean test MSE 0.003689 versus 0.010281 for the swept baseline, delta -0.006592, permutation p=0.0081. The trained-model mechanism signature was confirmed with mean absolute output gap 0.0699 below tolerance 0.25.",
  "metrics": {
    "baseline": "Test MSE mean=0.0102806917, std=0.0016466487; best lr=0.001, epochs=18.",
    "idea": "Test MSE mean=0.0036891116, std=0.0017249864; best lr=0.01, epochs=18; 8/8 paired wins.",
    "comparison": "delta_mean=-0.0065915801, permutation p=0.0081, verdict=idea better (significant).",
    "mechanism_signature": "observed_mean_abs_output_gap=0.0699216, observed_std=0.0113909, tolerance=0.25, confirmed=true"
  },
  "bench_report": {
    "bench_version": 1,
    "track": "particle_field_moments",
    "model": "custom_mlp",
    "metric_direction": "lower is better",
    "n_seeds": 8,
    "baseline": {
      "best_cfg": {
        "lr": 0.001,
        "epochs": 18
      },
      "sweep": [
        {
          "cfg": {
            "lr": 0.001,
            "epochs": 18
          },
          "mean": 0.0099995176
        },
        {
          "cfg": {
            "lr": 0.003,
            "epochs": 18
          },
          "mean": 0.0113862732
        },
        {
          "cfg": {
            "lr": 0.01,
            "epochs": 18
          },
          "mean": 0.0104406073
        }
      ],
      "full": {
        "mean": 0.0102806917,
        "std": 0.0016466487,
        "per_seed": [
          0.0105852205,
          0.0110057117,
          0.0112093491,
          0.0071977889,
          0.0093886713,
          0.0113911498,
          0.0127669293,
          0.0087007126
        ],
        "n": 8
      }
    },
    "idea": {
      "best_cfg": {
        "lr": 0.01,
        "epochs": 18
      },
      "mean": 0.0036891116,
      "std": 0.0017249864,
      "per_seed": [
        0.0027679566,
        0.0025160152,
        0.0025187477,
        0.0025933082,
        0.0037713193,
        0.0079902764,
        0.004178816,
        "0.0031764535"
      ],
      "n": 8
    },
    "comparison": {
      "delta_mean": -0.0065915801,
      "idea_wins": 8,
      "n_pairs": 8,
      "p_value": 0.0081,
      "verdict": "idea better (significant)",
      "system_worked": true
    },
    "mechanism_signature": {
      "prediction": "trained moment and direct systems should produce similar fields",
      "observed_mean_abs_output_gap": 0.0699215932,
      "observed_std_abs_output_gap": 0.0113908861,
      "tolerance": 0.25,
      "confirmed": true
    },
    "custom_track": {
      "name": "particle_field_moments",
      "file": "particle_moment_bench.py",
      "domain": "point-cloud spatial aggregation"
    }
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 particle_moment_bench.py",
  "files": [
    "particle_moment_bench.py",
    "bench_report.json"
  ],
  "limitations": "The custom task uses one query per example, fixed 32-particle clouds, a degree-2 polynomial target, and dense PyTorch moment computation. It does not test sparse multi-box hierarchies, non-polynomial compact-support kernels, GPU-specialized kernels, wall-clock inference speed, or large realistic point-cloud workloads.",
  "system_verdict": "worked",
  "practical_verdict": "helps",
  "mechanism_ok": 1,
  "system_judged": true
}