results.json

{
  "config": {
    "rank": 128,
    "alpha": 256,
    "lp_ratio": 16,
    "lr": 0.0002,
    "steps": 30,
    "batch": 2,
    "seqlen": 256
  },
  "gpu": "NVIDIA RTX 4500 Ada Generation",
  "arms": {
    "fp_AdamW": {
      "optimizer": "AdamW",
      "use_8bit": false,
      "trainable_params": 169869312,
      "optimizer_state_mb": 1358.96,
      "device_peak_gb": 17.043,
      "torch_peak_alloc_gb": 16.029,
      "final_loss": 11.1553,
      "first_loss": 14.4515,
      "loss_curve": [
        14.4515,
        14.3866,
        22.6908,
        14.4499,
        20.818,
        14.6913,
        13.2703,
        15.4108,
        14.6073,
        15.2515,
        14.4043,
        13.2842,
        13.0611,
        13.9808,
        12.6708,
        12.6474,
        12.7782,
        12.7836,
        12.6234,
        12.3701,
        12.0831,
        11.8733,
        11.7867,
        11.7898,
        11.7793,
        11.6492,
        11.4572,
        11.303,
        11.2097,
        11.1553
      ],
      "wall_s": 56.9
    },
    "PagedAdamW8bit": {
      "optimizer": "PagedAdamW8bit",
      "use_8bit": true,
      "trainable_params": 169869312,
      "optimizer_state_mb": 345.57,
      "device_peak_gb": 16.325,
      "torch_peak_alloc_gb": 14.629,
      "final_loss": 10.6386,
      "first_loss": 14.4515,
      "loss_curve": [
        14.4515,
        14.3805,
        22.6782,
        14.5225,
        21.7097,
        14.8528,
        13.8653,
        13.3947,
        16.8753,
        13.1295,
        13.1709,
        13.2769,
        13.0932,
        12.4483,
        12.2095,
        12.424,
        11.9753,
        11.8203,
        11.8146,
        11.7498,
        11.5926,
        11.3962,
        11.238,
        11.1691,
        11.1169,
        11.0083,
        10.8668,
        10.7681,
        10.7138,
        10.6386
      ],
      "wall_s": 7.0
    }
  },
  "error": null,
  "dtype": "torch.bfloat16",
  "model": "Qwen/Qwen3.5-4B"
}