Knieper Rollout Stability Metric / results.json

Failed on benchmark

Raw ⬇ ZIP
  1{
  2  "toy": {
  3    "scalar_boundary": {
  4      "H": 8,
  5      "predicted_boundary_abs_a": 1.0,
  6      "sweep_a": [
  7        0.8,
  8        0.95,
  9        0.99,
 10        1.0,
 11        1.01,
 12        1.05,
 13        1.2
 14      ],
 15      "observed_gain": [
 16        1.0,
 17        1.0,
 18        1.0,
 19        1.0,
 20        1.08285670562808,
 21        1.477455443789063,
 22        4.299816959999999
 23      ],
 24      "predicted_gain": [
 25        1.0,
 26        1.0,
 27        1.0,
 28        1.0,
 29        1.0828567056280802,
 30        1.477455443789063,
 31        4.2998169599999985
 32      ]
 33    },
 34    "transient_blind_spot": {
 35      "K": [
 36        0.25,
 37        0.75,
 38        1.0,
 39        2.0,
 40        4.0
 41      ],
 42      "predicted_DH_over_eps": [
 43        1.0,
 44        1.0,
 45        1.0,
 46        2.0,
 47        4.0
 48      ],
 49      "observed_DH_over_eps": [
 50        1.0,
 51        1.0,
 52        1.0,
 53        2.0,
 54        4.0
 55      ],
 56      "final_step_gain": [
 57        0.0,
 58        0.0,
 59        0.0,
 60        0.0,
 61        0.0
 62      ]
 63    },
 64    "nonnormal_scaling": {
 65      "rho": 0.8,
 66      "H": 5,
 67      "K": [
 68        0.0,
 69        0.5,
 70        1.0,
 71        2.0,
 72        4.0
 73      ],
 74      "observed_gain": [
 75        1.0,
 76        1.1028817525011467,
 77        2.088558392767605,
 78        4.1164290544111175,
 79        8.202233607987523
 80      ],
 81      "predicted_gain": [
 82        1.0,
 83        1.1028817525011465,
 84        2.088558392767605,
 85        4.1164290544111175,
 86        8.202233607987523
 87      ],
 88      "slope_observed_large_K": 2.028386244424679
 89    }
 90  },
 91  "gru": {
 92    "device": "cpu_fallback",
 93    "cuda_error": "RuntimeError('cuDNN error: CUDNN_STATUS_INTERNAL_ERROR_HOST_ALLOCATION_FAILED')",
 94    "result": {
 95      "device": "cpu",
 96      "baseline": {
 97        "loss": 0.9444323182106018,
 98        "median_G8": 0.5372055768966675
 99      },
100      "rollout_lambda_0.03": {
101        "loss": 0.9449787139892578,
102        "median_G8": 0.09716863930225372
103      }
104    }
105  }
106}