data/e4_gamma/EleutherAI--pythia-31m_random.json

{
  "model": "EleutherAI/pythia-31m",
  "theta": 10000,
  "gamma_pred": 0.99999561666838,
  "C_theory": 9.2103,
  "distances": [
    10,
    20,
    30,
    50,
    100,
    200,
    500,
    1000,
    2000
  ],
  "distances_fit": [
    30,
    50,
    100,
    200,
    500,
    1000,
    2000
  ],
  "d_min_fit": 30,
  "n_prompts_per_distance": 150,
  "seeds": [
    42,
    123,
    7
  ],
  "attn_by_distance": {
    "10": {
      "mean_across_seeds": 0.01301000003495978,
      "std_across_seeds": 0.00046688234450363445,
      "per_seed_means": [
        0.013054606715838114,
        0.01241719133220613,
        0.013558202056835096
      ]
    },
    "20": {
      "mean_across_seeds": 0.00693967725809974,
      "std_across_seeds": 0.0003111505620949703,
      "per_seed_means": [
        0.006937875649115691,
        0.006559501201457654,
        0.007321654923725873
      ]
    },
    "30": {
      "mean_across_seeds": 0.004015894493398567,
      "std_across_seeds": 0.00011863265749955042,
      "per_seed_means": [
        0.0040281298368548355,
        0.0038648689771071075,
        0.004154684666233758
      ]
    },
    "50": {
      "mean_across_seeds": 0.0036222275638202617,
      "std_across_seeds": 0.0001889761810972585,
      "per_seed_means": [
        0.003356990779672439,
        0.003726471992752825,
        0.0037832199190355217
      ]
    },
    "100": {
      "mean_across_seeds": 0.0010882257377185548,
      "std_across_seeds": 0.00012946031918837256,
      "per_seed_means": [
        0.0009092199558411569,
        0.0012110101386497262,
        0.0011444471186647812
      ]
    },
    "200": {
      "mean_across_seeds": 0.0005025619622918183,
      "std_across_seeds": 6.674565621401741e-05,
      "per_seed_means": [
        0.0004423195596852262,
        0.0005956165699171834,
        0.00046974975727304504
      ]
    },
    "500": {
      "mean_across_seeds": 0.00019388230073268966,
      "std_across_seeds": 9.190553129509277e-06,
      "per_seed_means": [
        0.0001877195072908459,
        0.00018705338045644262,
        0.00020687401445078043
      ]
    },
    "1000": {
      "mean_across_seeds": 2.1299731392648228e-05,
      "std_across_seeds": 7.105662382177325e-06,
      "per_seed_means": [
        2.9809543584254546e-05,
        2.167332172462011e-05,
        1.2416328869070034e-05
      ]
    },
    "2000": {
      "mean_across_seeds": 7.186259580025636e-06,
      "std_across_seeds": 8.041626579142191e-07,
      "per_seed_means": [
        6.1229779024548255e-06,
        7.368486014153556e-06,
        8.067314823468526e-06
      ]
    }
  },
  "fit_power_law": {
    "gamma": 1.5398244746231446,
    "log_A": 0.24504650180534032,
    "R2": 0.964259,
    "n_points": 7
  },
  "fit_exponential": {
    "lambda": 0.003151060152234357,
    "log_A": -6.354208754917211,
    "R2": 0.870219,
    "n_points": 7
  },
  "delta_R2_power_minus_exp": 0.094,
  "gamma_CI_95": {
    "lo": 1.1197173993826866,
    "hi": 1.769418762310705
  },
  "decision": "ANOMALY: long-context training effect"
}