{
  "mode": "static",
  "args": {
    "mode": "static",
    "seed": 53,
    "modulus": 11,
    "observe_mod": 4,
    "observe_prob": 0.3,
    "op_family": "full",
    "query_family": "all",
    "train_min_len": 1,
    "train_max_len": 6,
    "eval_lengths": "3,6,9,12",
    "eval_k": "0,1,2,3,6,9,12",
    "eval_query_types": "all",
    "state_dim": 256,
    "instr_dim": 128,
    "heads": 4,
    "compiler_layers": 2,
    "dropout": 0.0,
    "train_steps": 3000,
    "batch_size": 512,
    "eval_batch_size": 512,
    "eval_examples": 512,
    "lr": 0.001,
    "weight_decay": 0.0001,
    "probe_steps": 1500,
    "probe_batch_size": 512,
    "probe_lr": 0.001,
    "probe_weight_decay": 0.0001,
    "probe_dropout": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 200,
    "probe_log_every": 150,
    "output_dir": "experiments/dense_latent_query_executor/runs/control_static_mod11",
    "checkpoint_dir": "large_artifacts/dense_latent_query_executor/checkpoints/control_static_mod11"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 2.4122157096862793,
      "grad_norm": 0.6011383533477783,
      "query_target_mass": 0.44674748182296753,
      "query_target_nll": 2.341661214828491,
      "query_top1_on_support": 0.515625,
      "mean_query_support_size": 4.75390625
    },
    {
      "step": 200,
      "loss": 2.1805124282836914,
      "grad_norm": 0.41278791427612305,
      "query_target_mass": 0.5304515957832336,
      "query_target_nll": 2.1797358989715576,
      "query_top1_on_support": 0.5859375,
      "mean_query_support_size": 4.9140625
    },
    {
      "step": 400,
      "loss": 2.114121198654175,
      "grad_norm": 0.47501620650291443,
      "query_target_mass": 0.5815714001655579,
      "query_target_nll": 2.095879077911377,
      "query_top1_on_support": 0.58984375,
      "mean_query_support_size": 5.076171875
    },
    {
      "step": 600,
      "loss": 2.0262975692749023,
      "grad_norm": 0.5089429020881653,
      "query_target_mass": 0.5797173380851746,
      "query_target_nll": 2.014464855194092,
      "query_top1_on_support": 0.626953125,
      "mean_query_support_size": 4.62890625
    },
    {
      "step": 800,
      "loss": 2.1040117740631104,
      "grad_norm": 0.4536289572715759,
      "query_target_mass": 0.57242751121521,
      "query_target_nll": 2.0817160606384277,
      "query_top1_on_support": 0.615234375,
      "mean_query_support_size": 4.859375
    },
    {
      "step": 1000,
      "loss": 2.06329345703125,
      "grad_norm": 0.47344473004341125,
      "query_target_mass": 0.6171896457672119,
      "query_target_nll": 2.038789749145508,
      "query_top1_on_support": 0.642578125,
      "mean_query_support_size": 5.09375
    },
    {
      "step": 1200,
      "loss": 2.0077333450317383,
      "grad_norm": 0.45551812648773193,
      "query_target_mass": 0.6024959683418274,
      "query_target_nll": 2.013890266418457,
      "query_top1_on_support": 0.626953125,
      "mean_query_support_size": 4.83203125
    },
    {
      "step": 1400,
      "loss": 1.9674631357192993,
      "grad_norm": 0.47805699706077576,
      "query_target_mass": 0.5860326290130615,
      "query_target_nll": 1.9590160846710205,
      "query_top1_on_support": 0.615234375,
      "mean_query_support_size": 4.568359375
    },
    {
      "step": 1600,
      "loss": 1.9779319763183594,
      "grad_norm": 0.601224958896637,
      "query_target_mass": 0.6472043991088867,
      "query_target_nll": 1.9459905624389648,
      "query_top1_on_support": 0.689453125,
      "mean_query_support_size": 5.001953125
    },
    {
      "step": 1800,
      "loss": 1.8987460136413574,
      "grad_norm": 0.6242047548294067,
      "query_target_mass": 0.6869659423828125,
      "query_target_nll": 1.8813502788543701,
      "query_top1_on_support": 0.76171875,
      "mean_query_support_size": 5.134765625
    },
    {
      "step": 2000,
      "loss": 1.9166336059570312,
      "grad_norm": 0.6355232000350952,
      "query_target_mass": 0.6752363443374634,
      "query_target_nll": 1.9000873565673828,
      "query_top1_on_support": 0.724609375,
      "mean_query_support_size": 5.052734375
    },
    {
      "step": 2200,
      "loss": 1.8724716901779175,
      "grad_norm": 0.6236591935157776,
      "query_target_mass": 0.6719210743904114,
      "query_target_nll": 1.8543018102645874,
      "query_top1_on_support": 0.734375,
      "mean_query_support_size": 4.8671875
    },
    {
      "step": 2400,
      "loss": 1.7877010107040405,
      "grad_norm": 0.5686831474304199,
      "query_target_mass": 0.6771365404129028,
      "query_target_nll": 1.771895170211792,
      "query_top1_on_support": 0.763671875,
      "mean_query_support_size": 4.7109375
    },
    {
      "step": 2600,
      "loss": 1.7713732719421387,
      "grad_norm": 0.6246246099472046,
      "query_target_mass": 0.6912781000137329,
      "query_target_nll": 1.747495174407959,
      "query_top1_on_support": 0.76171875,
      "mean_query_support_size": 4.685546875
    },
    {
      "step": 2800,
      "loss": 1.735321044921875,
      "grad_norm": 0.6088602542877197,
      "query_target_mass": 0.7300446629524231,
      "query_target_nll": 1.7166216373443604,
      "query_top1_on_support": 0.798828125,
      "mean_query_support_size": 4.97265625
    },
    {
      "step": 3000,
      "loss": 1.7516005039215088,
      "grad_norm": 0.6509389281272888,
      "query_target_mass": 0.7354283928871155,
      "query_target_nll": 1.7398459911346436,
      "query_top1_on_support": 0.7890625,
      "mean_query_support_size": 4.916015625
    }
  ],
  "probe_train": [
    {
      "step": 1,
      "loss": 4.809675216674805,
      "grad_norm": 0.4555661976337433,
      "probe_belief_target_mass": 0.04726499319076538,
      "probe_belief_target_nll": 4.809675216674805,
      "probe_belief_top1_on_support": 0.048828125,
      "mean_belief_support_size": 5.681640625
    },
    {
      "step": 150,
      "loss": 2.742042064666748,
      "grad_norm": 0.4823080599308014,
      "probe_belief_target_mass": 0.488213449716568,
      "probe_belief_target_nll": 2.742042064666748,
      "probe_belief_top1_on_support": 0.58984375,
      "mean_belief_support_size": 5.501953125
    },
    {
      "step": 300,
      "loss": 2.867673397064209,
      "grad_norm": 0.5250405073165894,
      "probe_belief_target_mass": 0.47278329730033875,
      "probe_belief_target_nll": 2.867673397064209,
      "probe_belief_top1_on_support": 0.564453125,
      "mean_belief_support_size": 5.58984375
    },
    {
      "step": 450,
      "loss": 2.5946545600891113,
      "grad_norm": 0.5041904449462891,
      "probe_belief_target_mass": 0.5481995344161987,
      "probe_belief_target_nll": 2.5946545600891113,
      "probe_belief_top1_on_support": 0.64453125,
      "mean_belief_support_size": 5.4609375
    },
    {
      "step": 600,
      "loss": 2.73473858833313,
      "grad_norm": 0.575081467628479,
      "probe_belief_target_mass": 0.5196862816810608,
      "probe_belief_target_nll": 2.73473858833313,
      "probe_belief_top1_on_support": 0.60546875,
      "mean_belief_support_size": 5.701171875
    },
    {
      "step": 750,
      "loss": 2.774055004119873,
      "grad_norm": 0.5133222937583923,
      "probe_belief_target_mass": 0.5016529560089111,
      "probe_belief_target_nll": 2.774055004119873,
      "probe_belief_top1_on_support": 0.611328125,
      "mean_belief_support_size": 5.765625
    },
    {
      "step": 900,
      "loss": 2.659860849380493,
      "grad_norm": 0.5771536231040955,
      "probe_belief_target_mass": 0.5301756262779236,
      "probe_belief_target_nll": 2.659860849380493,
      "probe_belief_top1_on_support": 0.625,
      "mean_belief_support_size": 5.740234375
    },
    {
      "step": 1050,
      "loss": 2.621800422668457,
      "grad_norm": 0.5140261054039001,
      "probe_belief_target_mass": 0.5865390300750732,
      "probe_belief_target_nll": 2.621800422668457,
      "probe_belief_top1_on_support": 0.681640625,
      "mean_belief_support_size": 6.203125
    },
    {
      "step": 1200,
      "loss": 2.642996072769165,
      "grad_norm": 0.5422481894493103,
      "probe_belief_target_mass": 0.534953773021698,
      "probe_belief_target_nll": 2.642996072769165,
      "probe_belief_top1_on_support": 0.630859375,
      "mean_belief_support_size": 5.6953125
    },
    {
      "step": 1350,
      "loss": 2.6214089393615723,
      "grad_norm": 0.5449801683425903,
      "probe_belief_target_mass": 0.5570270419120789,
      "probe_belief_target_nll": 2.6214089393615723,
      "probe_belief_top1_on_support": 0.662109375,
      "mean_belief_support_size": 5.576171875
    },
    {
      "step": 1500,
      "loss": 2.563278913497925,
      "grad_norm": 0.5610576272010803,
      "probe_belief_target_mass": 0.5812165141105652,
      "probe_belief_target_nll": 2.563278913497925,
      "probe_belief_top1_on_support": 0.681640625,
      "mean_belief_support_size": 5.654296875
    }
  ],
  "eval": [
    {
      "step": 3000,
      "rows": [
        {
          "model": "static",
          "length": 3,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.7881993651390076,
          "query_target_nll": 1.6529829502105713,
          "query_top1_on_support": 0.849609375,
          "mean_query_support_size": 4.888671875,
          "probe_belief_target_mass": 0.5031310319900513,
          "probe_belief_target_nll": 2.731247901916504,
          "probe_belief_top1_on_support": 0.634765625,
          "mean_belief_support_size": 5.791015625
        },
        {
          "model": "static",
          "length": 3,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.7858917713165283,
          "query_target_nll": 1.639503836631775,
          "query_top1_on_support": 0.826171875,
          "mean_query_support_size": 4.830078125,
          "probe_belief_target_mass": 0.5320981740951538,
          "probe_belief_target_nll": 2.5884714126586914,
          "probe_belief_top1_on_support": 0.642578125,
          "mean_belief_support_size": 5.55859375
        },
        {
          "model": "static",
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.6925997734069824,
          "query_target_nll": 2.0496749877929688,
          "query_top1_on_support": 0.7578125,
          "mean_query_support_size": 5.654296875,
          "probe_belief_target_mass": 0.5750269293785095,
          "probe_belief_target_nll": 2.538818597793579,
          "probe_belief_top1_on_support": 0.6953125,
          "mean_belief_support_size": 5.87109375
        },
        {
          "model": "static",
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.6932682991027832,
          "query_target_nll": 1.7061562538146973,
          "query_top1_on_support": 0.75390625,
          "mean_query_support_size": 4.8515625,
          "probe_belief_target_mass": 0.5373903512954712,
          "probe_belief_target_nll": 2.703399419784546,
          "probe_belief_top1_on_support": 0.64453125,
          "mean_belief_support_size": 6.1484375
        },
        {
          "model": "static",
          "length": 6,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.5265558958053589,
          "query_target_nll": 1.8011047840118408,
          "query_top1_on_support": 0.564453125,
          "mean_query_support_size": 3.263671875,
          "probe_belief_target_mass": 0.25575459003448486,
          "probe_belief_target_nll": 3.2843198776245117,
          "probe_belief_top1_on_support": 0.337890625,
          "mean_belief_support_size": 3.919921875
        },
        {
          "model": "static",
          "length": 6,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.5019991397857666,
          "query_target_nll": 1.8350720405578613,
          "query_top1_on_support": 0.5703125,
          "mean_query_support_size": 3.142578125,
          "probe_belief_target_mass": 0.23573929071426392,
          "probe_belief_target_nll": 3.308133602142334,
          "probe_belief_top1_on_support": 0.310546875,
          "mean_belief_support_size": 3.673828125
        },
        {
          "model": "static",
          "length": 6,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.3925970792770386,
          "query_target_nll": 2.1884121894836426,
          "query_top1_on_support": 0.455078125,
          "mean_query_support_size": 3.419921875,
          "probe_belief_target_mass": 0.2488352656364441,
          "probe_belief_target_nll": 3.261833429336548,
          "probe_belief_top1_on_support": 0.310546875,
          "mean_belief_support_size": 3.583984375
        },
        {
          "model": "static",
          "length": 6,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.4258195161819458,
          "query_target_nll": 2.059234380722046,
          "query_top1_on_support": 0.4921875,
          "mean_query_support_size": 3.419921875,
          "probe_belief_target_mass": 0.2522426247596741,
          "probe_belief_target_nll": 3.169381856918335,
          "probe_belief_top1_on_support": 0.32421875,
          "mean_belief_support_size": 3.73046875
        },
        {
          "model": "static",
          "length": 9,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.27652817964553833,
          "query_target_nll": 3.070720672607422,
          "query_top1_on_support": 0.30078125,
          "mean_query_support_size": 2.224609375,
          "probe_belief_target_mass": 0.05032799392938614,
          "probe_belief_target_nll": 7.733750343322754,
          "probe_belief_top1_on_support": 0.052734375,
          "mean_belief_support_size": 2.44140625
        },
        {
          "model": "static",
          "length": 9,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.259814590215683,
          "query_target_nll": 3.349393844604492,
          "query_top1_on_support": 0.28515625,
          "mean_query_support_size": 2.19140625,
          "probe_belief_target_mass": 0.05054497346282005,
          "probe_belief_target_nll": 7.859610080718994,
          "probe_belief_top1_on_support": 0.05078125,
          "mean_belief_support_size": 2.490234375
        },
        {
          "model": "static",
          "length": 9,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.22875770926475525,
          "query_target_nll": 2.604599952697754,
          "query_top1_on_support": 0.228515625,
          "mean_query_support_size": 2.4921875,
          "probe_belief_target_mass": 0.041229166090488434,
          "probe_belief_target_nll": 7.872122287750244,
          "probe_belief_top1_on_support": 0.03515625,
          "mean_belief_support_size": 2.705078125
        },
        {
          "model": "static",
          "length": 9,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.23715537786483765,
          "query_target_nll": 2.967322826385498,
          "query_top1_on_support": 0.240234375,
          "mean_query_support_size": 2.46875,
          "probe_belief_target_mass": 0.04982956498861313,
          "probe_belief_target_nll": 7.5499725341796875,
          "probe_belief_top1_on_support": 0.056640625,
          "mean_belief_support_size": 2.6953125
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.21091260015964508,
          "query_target_nll": 3.372745990753174,
          "query_top1_on_support": 0.232421875,
          "mean_query_support_size": 1.888671875,
          "probe_belief_target_mass": 0.02584109827876091,
          "probe_belief_target_nll": 9.740568161010742,
          "probe_belief_top1_on_support": 0.029296875,
          "mean_belief_support_size": 2.05078125
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.1903809905052185,
          "query_target_nll": 3.5272111892700195,
          "query_top1_on_support": 0.177734375,
          "mean_query_support_size": 1.822265625,
          "probe_belief_target_mass": 0.0241951122879982,
          "probe_belief_target_nll": 9.627504348754883,
          "probe_belief_top1_on_support": 0.025390625,
          "mean_belief_support_size": 1.966796875
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.17289619147777557,
          "query_target_nll": 2.6487293243408203,
          "query_top1_on_support": 0.16015625,
          "mean_query_support_size": 1.974609375,
          "probe_belief_target_mass": 0.028109000995755196,
          "probe_belief_target_nll": 9.142465591430664,
          "probe_belief_top1_on_support": 0.033203125,
          "mean_belief_support_size": 2.1328125
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.17547917366027832,
          "query_target_nll": 3.6041269302368164,
          "query_top1_on_support": 0.1796875,
          "mean_query_support_size": 1.939453125,
          "probe_belief_target_mass": 0.02218581549823284,
          "probe_belief_target_nll": 9.476716995239258,
          "probe_belief_top1_on_support": 0.025390625,
          "mean_belief_support_size": 1.99609375
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/dense_latent_query_executor/checkpoints/control_static_mod11/checkpoint_final.pt"
  ]
}