final_adapter/metadata.json

{
  "experiment": "ssd-Qwen3.6-35B-A3B-r16-full-20260505-184155",
  "model": "Qwen/Qwen3.6-35B-A3B",
  "config": {
    "model_name": "Qwen/Qwen3.6-35B-A3B",
    "max_seq_length": 2048,
    "load_in_4bit": false,
    "load_in_8bit": false,
    "lora_r": 16,
    "lora_alpha": 16,
    "lora_dropout": 0.0,
    "lora_bias": "none",
    "use_rslora": false,
    "optim": "adamw_8bit",
    "batch_size": 4,
    "gradient_accumulation_steps": 8,
    "packing": false,
    "learning_rate": 0.0002,
    "lr_scheduler_type": "cosine",
    "warmup_steps": 0.06,
    "weight_decay": 0.01,
    "max_steps": 150,
    "save_steps": 50,
    "eval_steps": 25,
    "logging_steps": 5,
    "seed": 42,
    "experiment_name": "ssd-Qwen3.6-35B-A3B-r16-full-20260505-184155",
    "skip_eval": false,
    "smoke_test": false
  },
  "results": {
    "train_loss": 0.5228066476186116,
    "steps": 150,
    "epoch": null
  },
  "timings": {
    "model_load_sec": 31.11,
    "data_load_sec": 22.24,
    "lora_setup_sec": 9.57,
    "training_sec": 4606.66,
    "training_loss": 0.5228,
    "training_steps": 150,
    "save_sec": 1.65,
    "step_timing": {
      "total_steps": 150,
      "avg_step_sec": 29.347,
      "min_step_sec": 14.719,
      "max_step_sec": 33.735,
      "total_step_sec": 4402.09,
      "eval_count": 5,
      "total_eval_sec": 3821.16,
      "save_count": 3,
      "total_save_sec": 10.16,
      "peak_gpu_gb": 64.7
    },
    "total_sec": 4690.74,
    "total_min": 78.18,
    "total_params": 34679779968,
    "trainable_params": 19169280,
    "trainable_pct": 0.0553
  }
}