{
  "config": {
    "run_name": "default-seed2",
    "seed": 2,
    "quick": false,
    "device": "auto",
    "eval_every": 250,
    "task": {
      "n_values": 64,
      "length": 16
    },
    "splits": {
      "val_seed": 1000003,
      "val_size": 1024,
      "test_seed": 2000003,
      "test_size": 2048
    },
    "model": {
      "n_layers": 4,
      "d_model": 256,
      "n_heads": 8,
      "d_mlp": 1024,
      "mlp": "gelu",
      "norm": "layernorm",
      "dropout": 0.0
    },
    "optim": {
      "steps": 4000,
      "batch_size": 256,
      "lr": 0.001,
      "warmup_steps": 100,
      "min_lr_ratio": 0.1,
      "weight_decay": 0.1,
      "grad_clip": 1.0
    }
  },
  "model_config": {
    "vocab_size": 65,
    "n_layers": 4,
    "d_model": 256,
    "n_heads": 8,
    "d_mlp": 1024,
    "max_seq_len": 33,
    "mlp": "gelu",
    "norm": "layernorm",
    "dropout": 0.0
  },
  "parameters": 3201280,
  "data": {
    "task": "sorting",
    "version": 1,
    "n_values": 64,
    "length": 16,
    "val_seed": 1000003,
    "val_size": 1024,
    "test_seed": 2000003,
    "test_size": 2048,
    "train_seed": 2
  },
  "code": {
    "commit": "19276f75392bee9d5f6f61820395002a0746e349",
    "dirty": true,
    "snapshot": "source-snapshot.tar.gz"
  },
  "environment": {
    "python": "3.12.3",
    "torch": "2.15.0+cu132",
    "device": "cuda",
    "precision": "torch.float32",
    "uv_lock_sha256": "60eba889e8dcd90d00fdcc282a51fc42df3869f3ef3807b1946aaa3cfd5f62e2",
    "gpu": "NVIDIA GeForce RTX 3060",
    "cuda": "13.2"
  },
  "resumable": false,
  "status": "finished",
  "seconds": 646.7,
  "history": [
    {
      "step": 250,
      "train_loss": 0.08374407142400742,
      "val_loss": 0.08853542804718018,
      "val_token_accuracy": 0.97259521484375,
      "val_exact_match": 0.654296875,
      "grad_norm": 2.7647416591644287,
      "lr": 0.0009967189933441244
    },
    {
      "step": 500,
      "train_loss": 0.03046058677136898,
      "val_loss": 0.030035920441150665,
      "val_token_accuracy": 0.99041748046875,
      "val_exact_match": 0.8642578125,
      "grad_norm": 1.855226993560791,
      "lr": 0.0009768413988762155
    },
    {
      "step": 750,
      "train_loss": 0.019605081528425217,
      "val_loss": 0.02953985333442688,
      "val_token_accuracy": 0.99114990234375,
      "val_exact_match": 0.87109375,
      "grad_norm": 0.8727502226829529,
      "lr": 0.0009397114317029974
    },
    {
      "step": 1000,
      "train_loss": 0.010449330322444439,
      "val_loss": 0.01985304430127144,
      "val_token_accuracy": 0.99554443359375,
      "val_exact_match": 0.9326171875,
      "grad_norm": 0.843723714351654,
      "lr": 0.0008868298366769955
    },
    {
      "step": 1250,
      "train_loss": 0.004702113103121519,
      "val_loss": 0.008932323195040226,
      "val_token_accuracy": 0.9974365234375,
      "val_exact_match": 0.962890625,
      "grad_norm": 0.41519755125045776,
      "lr": 0.0008203340189070906
    },
    {
      "step": 1500,
      "train_loss": 0.00736136082559824,
      "val_loss": 0.004941404797136784,
      "val_token_accuracy": 0.9989013671875,
      "val_exact_match": 0.982421875,
      "grad_norm": 1.172151803970337,
      "lr": 0.0007429116526313744
    },
    {
      "step": 1750,
      "train_loss": 0.009208878502249718,
      "val_loss": 0.005312320776283741,
      "val_token_accuracy": 0.998779296875,
      "val_exact_match": 0.9814453125,
      "grad_norm": 1.4689639806747437,
      "lr": 0.0006576920489294011
    },
    {
      "step": 2000,
      "train_loss": 0.0014178939163684845,
      "val_loss": 0.004395550116896629,
      "val_token_accuracy": 0.9991455078125,
      "val_exact_match": 0.986328125,
      "grad_norm": 0.2403556853532791,
      "lr": 0.0005681196730492369
    },
    {
      "step": 2250,
      "train_loss": 0.0022070351988077164,
      "val_loss": 0.0005514943040907383,
      "val_token_accuracy": 0.99981689453125,
      "val_exact_match": 0.9970703125,
      "grad_norm": 0.19152404367923737,
      "lr": 0.00047781492361400794
    },
    {
      "step": 2500,
      "train_loss": 0.0019115328323096037,
      "val_loss": 0.0011586558539420366,
      "val_token_accuracy": 0.99957275390625,
      "val_exact_match": 0.9931640625,
      "grad_norm": 0.22928588092327118,
      "lr": 0.0003904278008308589
    },
    {
      "step": 2750,
      "train_loss": 0.003090714570134878,
      "val_loss": 0.001331779290921986,
      "val_token_accuracy": 0.9996337890625,
      "val_exact_match": 0.994140625,
      "grad_norm": 0.6994138956069946,
      "lr": 0.0003094903782424895
    },
    {
      "step": 3000,
      "train_loss": 0.0001012918510241434,
      "val_loss": 0.0001072997329174541,
      "val_token_accuracy": 1.0,
      "val_exact_match": 1.0,
      "grad_norm": 0.010654399171471596,
      "lr": 0.0002382740409206803
    },
    {
      "step": 3250,
      "train_loss": 3.054919943679124e-05,
      "val_loss": 0.004428497049957514,
      "val_token_accuracy": 0.99945068359375,
      "val_exact_match": 0.9912109375,
      "grad_norm": 0.004973436240106821,
      "lr": 0.00017965726034785466
    },
    {
      "step": 3500,
      "train_loss": 0.00010091940930578858,
      "val_loss": 0.0005630630766972899,
      "val_token_accuracy": 0.99981689453125,
      "val_exact_match": 0.9970703125,
      "grad_norm": 0.17502489686012268,
      "lr": 0.00013600925035352912
    },
    {
      "step": 3750,
      "train_loss": 1.1987899597443175e-06,
      "val_loss": 4.915940280625364e-06,
      "val_token_accuracy": 1.0,
      "val_exact_match": 1.0,
      "grad_norm": 6.335216312436387e-05,
      "lr": 0.00010909420658097956
    },
    {
      "step": 4000,
      "train_loss": 4.847031505050836e-06,
      "val_loss": 0.0006610961863771081,
      "val_token_accuracy": 0.99969482421875,
      "val_exact_match": 0.9951171875,
      "grad_norm": 0.0012014994863420725,
      "lr": 0.0001
    }
  ],
  "final": {
    "val_loss": 0.0006610961863771081,
    "val_token_accuracy": 0.99969482421875,
    "val_exact_match": 0.9951171875
  }
}