training_log.json

[
  {
    "step": 1,
    "loss": 1.0076,
    "scale": -0.0,
    "tfa_MiB": 2.5,
    "ctx_tokens": 131072,
    "min": 3.1
  },
  {
    "step": 2,
    "loss": 1.1605,
    "scale": -0.0001,
    "tfa_MiB": 2.5,
    "ctx_tokens": 262144,
    "min": 6.4
  },
  {
    "step": 3,
    "loss": 1.0958,
    "scale": -0.0001,
    "tfa_MiB": 2.5,
    "ctx_tokens": 393216,
    "min": 9.9
  },
  {
    "step": 4,
    "loss": 1.166,
    "scale": -0.0001,
    "tfa_MiB": 2.5,
    "ctx_tokens": 524288,
    "min": 13.2
  },
  {
    "step": 5,
    "loss": 1.3911,
    "scale": -0.0001,
    "tfa_MiB": 2.5,
    "ctx_tokens": 655360,
    "min": 18.0
  },
  {
    "step": 6,
    "loss": 1.1414,
    "scale": -0.0001,
    "tfa_MiB": 2.5,
    "ctx_tokens": 786432,
    "min": 22.5
  },
  {
    "step": 7,
    "loss": 1.4494,
    "scale": -0.0001,
    "tfa_MiB": 2.5,
    "ctx_tokens": 917504,
    "min": 27.6
  },
  {
    "step": 8,
    "loss": 1.8025,
    "scale": -0.0002,
    "tfa_MiB": 2.5,
    "ctx_tokens": 1048576,
    "min": 31.2
  },
  {
    "step": 9,
    "loss": 1.9003,
    "scale": -0.0002,
    "tfa_MiB": 2.5,
    "ctx_tokens": 1179648,
    "min": 35.3
  },
  {
    "step": 10,
    "loss": 1.8974,
    "scale": -0.0002,
    "tfa_MiB": 2.5,
    "ctx_tokens": 1310720,
    "min": 43.5
  }
]