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

(универсальный)

Ответ агента:

{
  "worked": true,
  "confidence": 8,
  "verdict": "Built randomized-shift Sobol Gaussian-noise minibatch sampling as the only training intervention, with an identical mlp_tiny baseline and a shared three-value learning-rate sweep. Across 8 paired seeds, QMC reduced test MSE from 0.009381 to 0.005704, with paired delta -0.003677 and permutation p=0.04775; the harness verdict is idea better (significant). The trained-model mechanism signature confirmed 69.6% lower gradient variance for QMC.",
  "metrics": {
    "baseline": "IID, best lr=0.006: mean test MSE 0.0093810774, std 0.0050767455.",
    "idea": "Randomized shifted Sobol, best lr=0.003: mean test MSE 0.0057039903, std 0.0020558108; paired delta=-0.0036770871, p=0.04775, 6/8 wins; gradient variance 1.63059e-05 vs IID 5.35511e-05, reduction=0.6955."
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 qmc_bench.py",
  "files": [
    "qmc_bench.py",
    "bench_report.json"
  ],
  "limitations": "The experiment uses a local custom smooth stochastic-regression track because built-in tracks do not expose an internal latent expectation. It uses one-dimensional noise, 18 epochs, batch size 32, and three learning rates; higher-dimensional latent vectors, Owen scrambling, wall-clock speed, larger models, and diffusion/VAE/CIFAR-10 transfer were not tested.",
  "bench_report": {
    "bench_version": 1,
    "track": "custom_stochastic_expectation",
    "model": "mlp_tiny",
    "metric_direction": "lower is better",
    "n_seeds": 8,
    "baseline": {
      "best_cfg": {
        "lr": 0.006
      },
      "sweep": [
        {
          "cfg": {
            "lr": 0.0015
          },
          "mean": 0.011956734815612435
        },
        {
          "cfg": {
            "lr": 0.003
          },
          "mean": 0.013272208627313375
        },
        {
          "cfg": {
            "lr": 0.006
          },
          "mean": 0.010816274618264288
        }
      ],
      "full": {
        "mean": 0.009381077397847548,
        "std": 0.005076745501359762,
        "per_seed": [
          0.021740078926086426,
          0.0081116808578372,
          0.009601366706192493,
          0.0038119719829410315,
          0.006551957223564386,
          0.006829647812992334,
          0.010959993116557598,
          0.007441922556608915
        ],
        "n": 8
      }
    },
    "idea": {
      "mean": 0.00570399034768343,
      "std": 0.002055810761713156,
      "per_seed": [
        0.006588147487491369,
        0.010348089039325714,
        0.0069997236132621765,
        0.00420125387609005,
        0.003918817266821861,
        0.00435355631634593,
        0.004861464258283377,
        0.00436087092384696
      ],
      "n": 8
    },
    "comparison": {
      "delta_mean": -0.0036770870501641184,
      "idea_wins": 6,
      "n_pairs": 8,
      "per_seed_diffs": [
        -0.015151931438595057,
        0.002236408181488514,
        -0.002601643092930317,
        0.0003892818931490183,
        -0.002633139956742525,
        -0.0024760914966464043,
        -0.006098528858274221,
        -0.0030810516327619553
      ],
      "p_value": 0.04775,
      "mde": 0.004391476661630811,
      "mde_rel_pct": 46.81207152856897,
      "verdict": "idea better (significant)",
      "system_worked": true
    },
    "mechanism_signature": {
      "track_structure": "explicit smooth Gaussian expectation in each minibatch",
      "gradient_variance": {
        "n": 32,
        "repetitions": 32,
        "predicted": "lower gradient variance for smooth base-noise integrand",
        "gradient_variance_iid": 5.355114990379661e-05,
        "gradient_variance_qmc": 1.6305946701322682e-05,
        "variance_reduction": 0.695507066970253,
        "confirmed": true
      }
    },
    "custom_track": {
      "name": "stochastic_expectation",
      "file": "qmc_bench.py",
      "domain": "stochastic_expectation"
    },
    "protocol": {
      "paired_seeds": [
        0,
        1,
        2,
        3,
        4,
        5,
        6,
        7
      ],
      "epochs": 18,
      "batch": 32,
      "lr_union_both_sides": [
        0.0015,
        0.003,
        0.006
      ],
      "model_parity": true
    }
  },
  "system_verdict": "worked",
  "practical_verdict": "helps",
  "mechanism_ok": 0,
  "system_judged": true
}