{
  "mode": "dense",
  "args": {
    "mode": "dense",
    "seed": 53,
    "modulus": 7,
    "observe_mod": 3,
    "observe_prob": 0.4,
    "op_family": "full",
    "query_family": "all",
    "train_min_len": 1,
    "train_max_len": 3,
    "eval_lengths": "2,3",
    "eval_k": "0,1,2,3",
    "eval_query_types": "all",
    "state_dim": 64,
    "instr_dim": 32,
    "heads": 4,
    "compiler_layers": 2,
    "dropout": 0.0,
    "train_steps": 2,
    "batch_size": 16,
    "eval_batch_size": 16,
    "eval_examples": 32,
    "lr": 0.001,
    "weight_decay": 0.0001,
    "probe_steps": 2,
    "probe_batch_size": 16,
    "probe_lr": 0.001,
    "probe_weight_decay": 0.0001,
    "probe_dropout": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 1,
    "probe_log_every": 1,
    "output_dir": "experiments/dense_latent_query_executor/runs/smoke_dense_mod7",
    "checkpoint_dir": "large_artifacts/dense_latent_query_executor/checkpoints/smoke_dense_mod7"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 1.9296256303787231,
      "grad_norm": 1.3355157375335693,
      "query_target_mass": 0.4111073911190033,
      "query_target_nll": 1.8895562887191772,
      "query_top1_on_support": 0.5625,
      "mean_query_support_size": 2.75
    },
    {
      "step": 2,
      "loss": 1.9333385229110718,
      "grad_norm": 1.2407244443893433,
      "query_target_mass": 0.5158318281173706,
      "query_target_nll": 1.91750967502594,
      "query_top1_on_support": 0.6875,
      "mean_query_support_size": 3.5625
    }
  ],
  "probe_train": [
    {
      "step": 1,
      "loss": 3.9025588035583496,
      "grad_norm": 0.6908218264579773,
      "probe_belief_target_mass": 0.10780919343233109,
      "probe_belief_target_nll": 3.9025588035583496,
      "probe_belief_top1_on_support": 0.09375,
      "mean_belief_support_size": 5.34375
    },
    {
      "step": 2,
      "loss": 3.864835023880005,
      "grad_norm": 0.6660990118980408,
      "probe_belief_target_mass": 0.11352226883172989,
      "probe_belief_target_nll": 3.864835023880005,
      "probe_belief_top1_on_support": 0.203125,
      "mean_belief_support_size": 5.390625
    }
  ],
  "eval": [
    {
      "step": 2,
      "rows": [
        {
          "model": "dense",
          "length": 2,
          "query_type": "A",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.501405194401741,
          "query_target_nll": 1.9566073417663574,
          "query_top1_on_support": 0.46875,
          "mean_query_support_size": 3.5,
          "probe_belief_target_mass": 0.07354598119854927,
          "probe_belief_target_nll": 3.8816674947738647,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 3.53125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.4995565712451935,
          "query_target_nll": 1.9604549407958984,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.5,
          "probe_belief_target_mass": 0.07296394556760788,
          "probe_belief_target_nll": 3.8832945823669434,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 3.53125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.4973133057355881,
          "query_target_nll": 1.9688852429389954,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.5,
          "probe_belief_target_mass": 0.07336843386292458,
          "probe_belief_target_nll": 3.883575201034546,
          "probe_belief_top1_on_support": 0.03125,
          "mean_belief_support_size": 3.53125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.4973133057355881,
          "query_target_nll": 1.9688852429389954,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.5,
          "probe_belief_target_mass": 0.07336843386292458,
          "probe_belief_target_nll": 3.883575201034546,
          "probe_belief_top1_on_support": 0.03125,
          "mean_belief_support_size": 3.53125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "B",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.5669002681970596,
          "query_target_nll": 1.9899840950965881,
          "query_top1_on_support": 0.53125,
          "mean_query_support_size": 4.0,
          "probe_belief_target_mass": 0.0875476561486721,
          "probe_belief_target_nll": 3.910546898841858,
          "probe_belief_top1_on_support": 0.09375,
          "mean_belief_support_size": 4.28125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "B",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.5710422843694687,
          "query_target_nll": 1.9680535793304443,
          "query_top1_on_support": 0.625,
          "mean_query_support_size": 4.0,
          "probe_belief_target_mass": 0.08675431832671165,
          "probe_belief_target_nll": 3.926242709159851,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.28125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "B",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.5724340230226517,
          "query_target_nll": 1.9590497612953186,
          "query_top1_on_support": 0.53125,
          "mean_query_support_size": 4.0,
          "probe_belief_target_mass": 0.08744879812002182,
          "probe_belief_target_nll": 3.9162920713424683,
          "probe_belief_top1_on_support": 0.1875,
          "mean_belief_support_size": 4.28125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "B",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.5724340230226517,
          "query_target_nll": 1.9590497612953186,
          "query_top1_on_support": 0.53125,
          "mean_query_support_size": 4.0,
          "probe_belief_target_mass": 0.08744879812002182,
          "probe_belief_target_nll": 3.9162920713424683,
          "probe_belief_top1_on_support": 0.1875,
          "mean_belief_support_size": 4.28125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.5342014133930206,
          "query_target_nll": 1.9562005400657654,
          "query_top1_on_support": 0.53125,
          "mean_query_support_size": 3.71875,
          "probe_belief_target_mass": 0.07758501917123795,
          "probe_belief_target_nll": 3.859814405441284,
          "probe_belief_top1_on_support": 0.09375,
          "mean_belief_support_size": 3.71875
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.5374526977539062,
          "query_target_nll": 1.9418547749519348,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 3.71875,
          "probe_belief_target_mass": 0.07675017416477203,
          "probe_belief_target_nll": 3.8730498552322388,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.71875
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.5358144044876099,
          "query_target_nll": 1.9414342641830444,
          "query_top1_on_support": 0.59375,
          "mean_query_support_size": 3.71875,
          "probe_belief_target_mass": 0.0765267014503479,
          "probe_belief_target_nll": 3.8743252754211426,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.71875
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.5358144044876099,
          "query_target_nll": 1.9414342641830444,
          "query_top1_on_support": 0.59375,
          "mean_query_support_size": 3.71875,
          "probe_belief_target_mass": 0.0765267014503479,
          "probe_belief_target_nll": 3.8743252754211426,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.71875
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.3344419300556183,
          "query_target_nll": 1.9584493041038513,
          "query_top1_on_support": 0.34375,
          "mean_query_support_size": 2.34375,
          "probe_belief_target_mass": 0.07834458351135254,
          "probe_belief_target_nll": 3.884928584098816,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.78125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.3340407758951187,
          "query_target_nll": 1.95391184091568,
          "query_top1_on_support": 0.375,
          "mean_query_support_size": 2.34375,
          "probe_belief_target_mass": 0.07944246754050255,
          "probe_belief_target_nll": 3.8784362077713013,
          "probe_belief_top1_on_support": 0.1875,
          "mean_belief_support_size": 3.78125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.33313414454460144,
          "query_target_nll": 1.9624669551849365,
          "query_top1_on_support": 0.3125,
          "mean_query_support_size": 2.34375,
          "probe_belief_target_mass": 0.07948539778590202,
          "probe_belief_target_nll": 3.8719429969787598,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.78125
        },
        {
          "model": "dense",
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.33313414454460144,
          "query_target_nll": 1.9624669551849365,
          "query_top1_on_support": 0.3125,
          "mean_query_support_size": 2.34375,
          "probe_belief_target_mass": 0.07948539778590202,
          "probe_belief_target_nll": 3.8719429969787598,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.78125
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.4544233977794647,
          "query_target_nll": 1.9674387574195862,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.07179101929068565,
          "probe_belief_target_nll": 3.9285932779312134,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.625
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.4548824280500412,
          "query_target_nll": 1.9575430154800415,
          "query_top1_on_support": 0.375,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.072972122579813,
          "probe_belief_target_nll": 3.9064849615097046,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.625
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.4558599144220352,
          "query_target_nll": 1.9516295194625854,
          "query_top1_on_support": 0.34375,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.0728050097823143,
          "probe_belief_target_nll": 3.8982290029525757,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 3.625
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.45892369747161865,
          "query_target_nll": 1.9418688416481018,
          "query_top1_on_support": 0.40625,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.07285216450691223,
          "probe_belief_target_nll": 3.9041858911514282,
          "probe_belief_top1_on_support": 0.09375,
          "mean_belief_support_size": 3.625
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "B",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.3468993008136749,
          "query_target_nll": 1.9632219076156616,
          "query_top1_on_support": 0.21875,
          "mean_query_support_size": 2.4375,
          "probe_belief_target_mass": 0.059621961787343025,
          "probe_belief_target_nll": 3.9307353496551514,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 2.9375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "B",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.34664005041122437,
          "query_target_nll": 1.9586549997329712,
          "query_top1_on_support": 0.34375,
          "mean_query_support_size": 2.4375,
          "probe_belief_target_mass": 0.05819578282535076,
          "probe_belief_target_nll": 3.948196291923523,
          "probe_belief_top1_on_support": 0.03125,
          "mean_belief_support_size": 2.9375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "B",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.35014136135578156,
          "query_target_nll": 1.953277826309204,
          "query_top1_on_support": 0.3125,
          "mean_query_support_size": 2.4375,
          "probe_belief_target_mass": 0.05762409046292305,
          "probe_belief_target_nll": 3.955756664276123,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 2.9375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "B",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.3530256450176239,
          "query_target_nll": 1.9511620998382568,
          "query_top1_on_support": 0.40625,
          "mean_query_support_size": 2.4375,
          "probe_belief_target_mass": 0.05830395594239235,
          "probe_belief_target_nll": 3.9384490251541138,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 2.9375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.697592169046402,
          "query_target_nll": 1.967762529850006,
          "query_top1_on_support": 0.65625,
          "mean_query_support_size": 4.90625,
          "probe_belief_target_mass": 0.10530712828040123,
          "probe_belief_target_nll": 3.9055930376052856,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 5.09375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.7020504772663116,
          "query_target_nll": 1.9518537521362305,
          "query_top1_on_support": 0.6875,
          "mean_query_support_size": 4.90625,
          "probe_belief_target_mass": 0.1034930944442749,
          "probe_belief_target_nll": 3.9164992570877075,
          "probe_belief_top1_on_support": 0.09375,
          "mean_belief_support_size": 5.09375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.7050279974937439,
          "query_target_nll": 1.9402191638946533,
          "query_top1_on_support": 0.71875,
          "mean_query_support_size": 4.90625,
          "probe_belief_target_mass": 0.1015002503991127,
          "probe_belief_target_nll": 3.931133270263672,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 5.09375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.7044085562229156,
          "query_target_nll": 1.9408015012741089,
          "query_top1_on_support": 0.78125,
          "mean_query_support_size": 4.90625,
          "probe_belief_target_mass": 0.1012490950524807,
          "probe_belief_target_nll": 3.9286582469940186,
          "probe_belief_top1_on_support": 0.03125,
          "mean_belief_support_size": 5.09375
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 0,
          "n": 32,
          "query_target_mass": 0.32857733219861984,
          "query_target_nll": 1.9514922499656677,
          "query_top1_on_support": 0.34375,
          "mean_query_support_size": 2.25,
          "probe_belief_target_mass": 0.05979788303375244,
          "probe_belief_target_nll": 3.8847177028656006,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 2.875
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 1,
          "n": 32,
          "query_target_mass": 0.3205243647098541,
          "query_target_nll": 1.9776424169540405,
          "query_top1_on_support": 0.25,
          "mean_query_support_size": 2.25,
          "probe_belief_target_mass": 0.05973221734166145,
          "probe_belief_target_nll": 3.873525857925415,
          "probe_belief_top1_on_support": 0.03125,
          "mean_belief_support_size": 2.875
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 2,
          "n": 32,
          "query_target_mass": 0.3168221414089203,
          "query_target_nll": 1.9979665279388428,
          "query_top1_on_support": 0.25,
          "mean_query_support_size": 2.25,
          "probe_belief_target_mass": 0.05971204861998558,
          "probe_belief_target_nll": 3.8798422813415527,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 2.875
        },
        {
          "model": "dense",
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 3,
          "n": 32,
          "query_target_mass": 0.3135398328304291,
          "query_target_nll": 2.0067144632339478,
          "query_top1_on_support": 0.21875,
          "mean_query_support_size": 2.25,
          "probe_belief_target_mass": 0.0597262866795063,
          "probe_belief_target_nll": 3.88383948802948,
          "probe_belief_top1_on_support": 0.03125,
          "mean_belief_support_size": 2.875
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/dense_latent_query_executor/checkpoints/smoke_dense_mod7/checkpoint_final.pt"
  ]
}