checkpoint-1/train_state.json

{
  "step": 1,
  "losses": [
    2.1343601644039154
  ],
  "dataset_offset": 0,
  "config": {
    "king_dir": "/root/king_165_hf",
    "optimizer": "muon_adamw",
    "lr": 5e-05,
    "muon_lr": 0.0001,
    "lora_r": 128,
    "steps": 5000,
    "grad_accum": 4,
    "seq_len": 2048,
    "unfreeze_head": false,
    "source_dirs": [
      "/root/datasets/finewebedu"
    ],
    "source_weights": [
      1.0
    ]
  }
}