outputs/charlm_summary.json

{
  "final_loss": {
    "ref_s0": 1.8969424962997437,
    "ref_s1": 1.8856981992721558,
    "flash_s0": 1.8889731168746948,
    "linear_s0": NaN
  },
  "mean_abs_delta_2nd_half": {
    "flash_vs_ref (method effect)": 0.005471451282501221,
    "seed_vs_ref (seed noise yardstick)": 0.008417630672454834
  },
  "linear_diverged": true,
  "nmse_median": {
    "m_companded": 2.5267644162517564e-05,
    "m_linear": 2.8218751714688336e-05,
    "v_companded": 9.618521817889085e-06,
    "v_linear": 0.7462301234211456
  },
  "nmse_improvement_x": {
    "momentum": 1.1167939335060169,
    "variance": 77582.62002725445
  },
  "config": {
    "ctx": 256,
    "batch": 32,
    "steps": 500,
    "lr": 0.001,
    "vocab": 65
  }
}