{
  "step": 2000,
  "val_loss": 1.0681076049804688,
  "model_config": {
    "sequence_len": 7180,
    "vocab_size": 1542,
    "n_layer": 20,
    "n_head": 10,
    "n_kv_head": 10,
    "n_embd": 1280,
    "window_pattern": "L"
  },
  "user_config": {
    "data_dir": "/data/sat2map/sat2map_g16_t12_target64_k1024",
    "run": "sat2map_h100_8gpu_v2",
    "device_type": "",
    "depth": 20,
    "aspect_ratio": 64,
    "head_dim": 128,
    "window_pattern": "L",
    "num_iterations": 2000,
    "device_batch_size": 1,
    "grad_accum_steps": 8,
    "total_batch_size": 0,
    "embedding_lr": 0.3,
    "unembedding_lr": 0.004,
    "weight_decay": 0.2,
    "matrix_lr": 0.02,
    "scalar_lr": 0.5,
    "adam_beta1": 0.8,
    "adam_beta2": 0.95,
    "warmup_ratio": 0.01,
    "warmdown_ratio": 0.4,
    "final_lr_frac": 0.0,
    "resume_from_step": -1,
    "eval_every": 200,
    "eval_steps": 50,
    "save_every": 500,
    "model_tag": null,
    "init_from_group": "",
    "init_from_tag": "",
    "init_from_step": -1
  },
  "dataset_meta": {
    "sequence_len": 7180,
    "sequence_tokens_len": 7181,
    "num_timesteps": 12,
    "timestep_strategy": "uniform",
    "context_grid_size": 16,
    "context_patch_size": 8,
    "target_grid_size": 64,
    "pad_id": 0,
    "mask_id": 1,
    "ignore_index": -1,
    "token_spec": {
      "s2_base": 2,
      "s2_k": 1024,
      "time_sep_id": 1026,
      "task_wc_id": 1027,
      "task_cdl_id": 1028,
      "bos_id": 1029,
      "wc_base": 1030,
      "cdl_base": 1286,
      "vocab_size": 1542
    },
    "task_id_to_name": {
      "1": "worldcover",
      "2": "cdl"
    }
  },
  "effective_total_batch_size": 459520,
  "grad_accum_steps": 8,
  "dataloader_state_dict": {
    "epoch": 14,
    "rank": 0
  },
  "loop_state": {
    "min_val_loss": 0.27583298087120056,
    "smooth_train_loss": 0.008957206082207212,
    "total_training_time": 1468.8923184871674
  }
}