{
  "mode": "static",
  "args": {
    "mode": "static",
    "seed": 53,
    "modulus": 31,
    "observe_mod": 5,
    "observe_prob": 0.3,
    "op_family": "full",
    "query_family": "all",
    "train_min_len": 1,
    "train_max_len": 8,
    "eval_lengths": "4,8,12,16,24",
    "eval_k": "0,1,2,4,8,12,16,24",
    "eval_query_types": "all",
    "state_dim": 256,
    "instr_dim": 128,
    "heads": 4,
    "compiler_layers": 2,
    "dropout": 0.0,
    "train_steps": 5000,
    "batch_size": 512,
    "eval_batch_size": 512,
    "eval_examples": 512,
    "lr": 0.001,
    "weight_decay": 0.0001,
    "probe_steps": 2000,
    "probe_batch_size": 512,
    "probe_lr": 0.001,
    "probe_weight_decay": 0.0001,
    "probe_dropout": 0.0,
    "max_grad_norm": 1.0,
    "log_every": 250,
    "probe_log_every": 200,
    "output_dir": "experiments/dense_latent_query_executor/runs/control_static_mod31",
    "checkpoint_dir": "large_artifacts/dense_latent_query_executor/checkpoints/control_static_mod31"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 3.454900026321411,
      "grad_norm": 0.6235178709030151,
      "query_target_mass": 0.3817371129989624,
      "query_target_nll": 3.381915330886841,
      "query_top1_on_support": 0.4765625,
      "mean_query_support_size": 11.65234375
    },
    {
      "step": 250,
      "loss": 3.270390033721924,
      "grad_norm": 0.41170617938041687,
      "query_target_mass": 0.4333716034889221,
      "query_target_nll": 3.257544994354248,
      "query_top1_on_support": 0.451171875,
      "mean_query_support_size": 11.8671875
    },
    {
      "step": 500,
      "loss": 3.1418046951293945,
      "grad_norm": 0.46448275446891785,
      "query_target_mass": 0.4239945709705353,
      "query_target_nll": 3.125290870666504,
      "query_top1_on_support": 0.47265625,
      "mean_query_support_size": 10.365234375
    },
    {
      "step": 750,
      "loss": 3.1756160259246826,
      "grad_norm": 0.4355730712413788,
      "query_target_mass": 0.4271523952484131,
      "query_target_nll": 3.1817805767059326,
      "query_top1_on_support": 0.462890625,
      "mean_query_support_size": 10.890625
    },
    {
      "step": 1000,
      "loss": 3.101102828979492,
      "grad_norm": 0.42332708835601807,
      "query_target_mass": 0.5002758502960205,
      "query_target_nll": 3.0692052841186523,
      "query_top1_on_support": 0.533203125,
      "mean_query_support_size": 12.0859375
    },
    {
      "step": 1250,
      "loss": 3.0569663047790527,
      "grad_norm": 0.46845629811286926,
      "query_target_mass": 0.4879835247993469,
      "query_target_nll": 3.0382437705993652,
      "query_top1_on_support": 0.53125,
      "mean_query_support_size": 11.146484375
    },
    {
      "step": 1500,
      "loss": 3.071761131286621,
      "grad_norm": 0.5005123615264893,
      "query_target_mass": 0.476243793964386,
      "query_target_nll": 3.06986927986145,
      "query_top1_on_support": 0.482421875,
      "mean_query_support_size": 11.0859375
    },
    {
      "step": 1750,
      "loss": 3.019573211669922,
      "grad_norm": 0.48671093583106995,
      "query_target_mass": 0.5141888856887817,
      "query_target_nll": 3.001786947250366,
      "query_top1_on_support": 0.568359375,
      "mean_query_support_size": 11.837890625
    },
    {
      "step": 2000,
      "loss": 3.0694451332092285,
      "grad_norm": 0.4688921570777893,
      "query_target_mass": 0.46449312567710876,
      "query_target_nll": 3.0512290000915527,
      "query_top1_on_support": 0.5,
      "mean_query_support_size": 10.689453125
    },
    {
      "step": 2250,
      "loss": 3.0177948474884033,
      "grad_norm": 0.4545142352581024,
      "query_target_mass": 0.49415355920791626,
      "query_target_nll": 2.9890544414520264,
      "query_top1_on_support": 0.525390625,
      "mean_query_support_size": 11.013671875
    },
    {
      "step": 2500,
      "loss": 3.055302143096924,
      "grad_norm": 0.49353209137916565,
      "query_target_mass": 0.49845993518829346,
      "query_target_nll": 3.037357807159424,
      "query_top1_on_support": 0.529296875,
      "mean_query_support_size": 11.212890625
    },
    {
      "step": 2750,
      "loss": 3.025851249694824,
      "grad_norm": 0.4507175385951996,
      "query_target_mass": 0.4991079568862915,
      "query_target_nll": 3.025424003601074,
      "query_top1_on_support": 0.544921875,
      "mean_query_support_size": 11.51953125
    },
    {
      "step": 3000,
      "loss": 2.968107223510742,
      "grad_norm": 0.45438721776008606,
      "query_target_mass": 0.5091122388839722,
      "query_target_nll": 2.96427059173584,
      "query_top1_on_support": 0.544921875,
      "mean_query_support_size": 10.9296875
    },
    {
      "step": 3250,
      "loss": 3.0218374729156494,
      "grad_norm": 0.4227367639541626,
      "query_target_mass": 0.5100937485694885,
      "query_target_nll": 2.996630907058716,
      "query_top1_on_support": 0.5625,
      "mean_query_support_size": 11.373046875
    },
    {
      "step": 3500,
      "loss": 2.9786174297332764,
      "grad_norm": 0.44948485493659973,
      "query_target_mass": 0.48938658833503723,
      "query_target_nll": 2.96079683303833,
      "query_top1_on_support": 0.509765625,
      "mean_query_support_size": 10.720703125
    },
    {
      "step": 3750,
      "loss": 2.9792628288269043,
      "grad_norm": 0.4823112487792969,
      "query_target_mass": 0.5060423612594604,
      "query_target_nll": 2.972806453704834,
      "query_top1_on_support": 0.533203125,
      "mean_query_support_size": 11.16015625
    },
    {
      "step": 4000,
      "loss": 2.9264073371887207,
      "grad_norm": 0.4091070890426636,
      "query_target_mass": 0.5147651433944702,
      "query_target_nll": 2.918344497680664,
      "query_top1_on_support": 0.55078125,
      "mean_query_support_size": 11.09765625
    },
    {
      "step": 4250,
      "loss": 2.9747414588928223,
      "grad_norm": 0.4590487778186798,
      "query_target_mass": 0.5208368301391602,
      "query_target_nll": 2.9581451416015625,
      "query_top1_on_support": 0.560546875,
      "mean_query_support_size": 11.41796875
    },
    {
      "step": 4500,
      "loss": 2.935044288635254,
      "grad_norm": 0.4725484549999237,
      "query_target_mass": 0.5027673244476318,
      "query_target_nll": 2.933720588684082,
      "query_top1_on_support": 0.5625,
      "mean_query_support_size": 10.8828125
    },
    {
      "step": 4750,
      "loss": 2.889552593231201,
      "grad_norm": 0.504491925239563,
      "query_target_mass": 0.5330871343612671,
      "query_target_nll": 2.865330457687378,
      "query_top1_on_support": 0.591796875,
      "mean_query_support_size": 11.30859375
    },
    {
      "step": 5000,
      "loss": 2.875152111053467,
      "grad_norm": 0.4838833510875702,
      "query_target_mass": 0.5359949469566345,
      "query_target_nll": 2.8596136569976807,
      "query_top1_on_support": 0.58984375,
      "mean_query_support_size": 10.8828125
    }
  ],
  "probe_train": [
    {
      "step": 1,
      "loss": 6.886910438537598,
      "grad_norm": 0.270370751619339,
      "probe_belief_target_mass": 0.01393265463411808,
      "probe_belief_target_nll": 6.886910438537598,
      "probe_belief_top1_on_support": 0.015625,
      "mean_belief_support_size": 13.36328125
    },
    {
      "step": 200,
      "loss": 5.334155082702637,
      "grad_norm": 0.46133095026016235,
      "probe_belief_target_mass": 0.14614591002464294,
      "probe_belief_target_nll": 5.334155082702637,
      "probe_belief_top1_on_support": 0.201171875,
      "mean_belief_support_size": 14.09375
    },
    {
      "step": 400,
      "loss": 5.155239105224609,
      "grad_norm": 0.450440913438797,
      "probe_belief_target_mass": 0.18381455540657043,
      "probe_belief_target_nll": 5.155239105224609,
      "probe_belief_top1_on_support": 0.232421875,
      "mean_belief_support_size": 14.205078125
    },
    {
      "step": 600,
      "loss": 5.231989860534668,
      "grad_norm": 0.41234225034713745,
      "probe_belief_target_mass": 0.1816718429327011,
      "probe_belief_target_nll": 5.231989860534668,
      "probe_belief_top1_on_support": 0.232421875,
      "mean_belief_support_size": 14.392578125
    },
    {
      "step": 800,
      "loss": 5.068723678588867,
      "grad_norm": 0.4272572100162506,
      "probe_belief_target_mass": 0.19688791036605835,
      "probe_belief_target_nll": 5.068723678588867,
      "probe_belief_top1_on_support": 0.240234375,
      "mean_belief_support_size": 13.19921875
    },
    {
      "step": 1000,
      "loss": 4.9685258865356445,
      "grad_norm": 0.4189678132534027,
      "probe_belief_target_mass": 0.20767581462860107,
      "probe_belief_target_nll": 4.9685258865356445,
      "probe_belief_top1_on_support": 0.291015625,
      "mean_belief_support_size": 14.169921875
    },
    {
      "step": 1200,
      "loss": 5.089025974273682,
      "grad_norm": 0.4074699282646179,
      "probe_belief_target_mass": 0.187615305185318,
      "probe_belief_target_nll": 5.089025974273682,
      "probe_belief_top1_on_support": 0.248046875,
      "mean_belief_support_size": 12.779296875
    },
    {
      "step": 1400,
      "loss": 4.9978837966918945,
      "grad_norm": 0.41879042983055115,
      "probe_belief_target_mass": 0.21971815824508667,
      "probe_belief_target_nll": 4.9978837966918945,
      "probe_belief_top1_on_support": 0.271484375,
      "mean_belief_support_size": 13.1640625
    },
    {
      "step": 1600,
      "loss": 4.986542701721191,
      "grad_norm": 0.44860661029815674,
      "probe_belief_target_mass": 0.20729516446590424,
      "probe_belief_target_nll": 4.986542701721191,
      "probe_belief_top1_on_support": 0.25390625,
      "mean_belief_support_size": 12.736328125
    },
    {
      "step": 1800,
      "loss": 5.035330295562744,
      "grad_norm": 0.3907996118068695,
      "probe_belief_target_mass": 0.19604921340942383,
      "probe_belief_target_nll": 5.035330295562744,
      "probe_belief_top1_on_support": 0.236328125,
      "mean_belief_support_size": 12.72265625
    },
    {
      "step": 2000,
      "loss": 4.983997344970703,
      "grad_norm": 0.41547444462776184,
      "probe_belief_target_mass": 0.21810656785964966,
      "probe_belief_target_nll": 4.983997344970703,
      "probe_belief_top1_on_support": 0.283203125,
      "mean_belief_support_size": 13.974609375
    }
  ],
  "eval": [
    {
      "step": 5000,
      "rows": [
        {
          "model": "static",
          "length": 4,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.5597519874572754,
          "query_target_nll": 2.955376625061035,
          "query_top1_on_support": 0.58203125,
          "mean_query_support_size": 10.779296875,
          "probe_belief_target_mass": 0.13676893711090088,
          "probe_belief_target_nll": 5.453972339630127,
          "probe_belief_top1_on_support": 0.181640625,
          "mean_belief_support_size": 13.26171875
        },
        {
          "model": "static",
          "length": 4,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.5920279026031494,
          "query_target_nll": 2.8322699069976807,
          "query_top1_on_support": 0.642578125,
          "mean_query_support_size": 10.986328125,
          "probe_belief_target_mass": 0.1493833363056183,
          "probe_belief_target_nll": 5.419735431671143,
          "probe_belief_top1_on_support": 0.19921875,
          "mean_belief_support_size": 13.19140625
        },
        {
          "model": "static",
          "length": 4,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.47140976786613464,
          "query_target_nll": 3.271993398666382,
          "query_top1_on_support": 0.525390625,
          "mean_query_support_size": 12.76171875,
          "probe_belief_target_mass": 0.11280353367328644,
          "probe_belief_target_nll": 5.557967185974121,
          "probe_belief_top1_on_support": 0.1484375,
          "mean_belief_support_size": 13.453125
        },
        {
          "model": "static",
          "length": 4,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.3920789361000061,
          "query_target_nll": 3.1666879653930664,
          "query_top1_on_support": 0.443359375,
          "mean_query_support_size": 9.9765625,
          "probe_belief_target_mass": 0.12184768915176392,
          "probe_belief_target_nll": 5.561555862426758,
          "probe_belief_top1_on_support": 0.162109375,
          "mean_belief_support_size": 12.560546875
        },
        {
          "model": "static",
          "length": 8,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.30215466022491455,
          "query_target_nll": 2.921665668487549,
          "query_top1_on_support": 0.306640625,
          "mean_query_support_size": 5.6015625,
          "probe_belief_target_mass": 0.03660156577825546,
          "probe_belief_target_nll": 5.697853088378906,
          "probe_belief_top1_on_support": 0.05078125,
          "mean_belief_support_size": 6.453125
        },
        {
          "model": "static",
          "length": 8,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.31473103165626526,
          "query_target_nll": 2.906665325164795,
          "query_top1_on_support": 0.330078125,
          "mean_query_support_size": 5.61328125,
          "probe_belief_target_mass": 0.03644798696041107,
          "probe_belief_target_nll": 5.664716720581055,
          "probe_belief_top1_on_support": 0.04296875,
          "mean_belief_support_size": 6.802734375
        },
        {
          "model": "static",
          "length": 8,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.23699167370796204,
          "query_target_nll": 3.2723464965820312,
          "query_top1_on_support": 0.2734375,
          "mean_query_support_size": 6.255859375,
          "probe_belief_target_mass": 0.04055093973875046,
          "probe_belief_target_nll": 5.626262664794922,
          "probe_belief_top1_on_support": 0.05078125,
          "mean_belief_support_size": 6.951171875
        },
        {
          "model": "static",
          "length": 8,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.22030764818191528,
          "query_target_nll": 3.251887321472168,
          "query_top1_on_support": 0.25390625,
          "mean_query_support_size": 5.470703125,
          "probe_belief_target_mass": 0.03292132541537285,
          "probe_belief_target_nll": 5.70545768737793,
          "probe_belief_top1_on_support": 0.037109375,
          "mean_belief_support_size": 6.341796875
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.13087917864322662,
          "query_target_nll": 4.290501117706299,
          "query_top1_on_support": 0.11328125,
          "mean_query_support_size": 3.208984375,
          "probe_belief_target_mass": 0.007924235425889492,
          "probe_belief_target_nll": 8.505693435668945,
          "probe_belief_top1_on_support": 0.005859375,
          "mean_belief_support_size": 3.849609375
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.153312087059021,
          "query_target_nll": 4.030825614929199,
          "query_top1_on_support": 0.115234375,
          "mean_query_support_size": 3.640625,
          "probe_belief_target_mass": 0.009892504662275314,
          "probe_belief_target_nll": 8.379608154296875,
          "probe_belief_top1_on_support": 0.0078125,
          "mean_belief_support_size": 4.38671875
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.1259702444076538,
          "query_target_nll": 3.5394530296325684,
          "query_top1_on_support": 0.111328125,
          "mean_query_support_size": 3.802734375,
          "probe_belief_target_mass": 0.0072226254269480705,
          "probe_belief_target_nll": 8.40391731262207,
          "probe_belief_top1_on_support": 0.00390625,
          "mean_belief_support_size": 4.041015625
        },
        {
          "model": "static",
          "length": 12,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.1322055459022522,
          "query_target_nll": 3.6925082206726074,
          "query_top1_on_support": 0.130859375,
          "mean_query_support_size": 3.9609375,
          "probe_belief_target_mass": 0.008098677732050419,
          "probe_belief_target_nll": 8.332406997680664,
          "probe_belief_top1_on_support": 0.001953125,
          "mean_belief_support_size": 4.236328125
        },
        {
          "model": "static",
          "length": 16,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.09329064190387726,
          "query_target_nll": 4.55894136428833,
          "query_top1_on_support": 0.099609375,
          "mean_query_support_size": 2.619140625,
          "probe_belief_target_mass": 0.0036422228440642357,
          "probe_belief_target_nll": 9.237142562866211,
          "probe_belief_top1_on_support": 0.001953125,
          "mean_belief_support_size": 3.0
        },
        {
          "model": "static",
          "length": 16,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.08856208622455597,
          "query_target_nll": 4.398288726806641,
          "query_top1_on_support": 0.095703125,
          "mean_query_support_size": 2.498046875,
          "probe_belief_target_mass": 0.003846000414341688,
          "probe_belief_target_nll": 9.34157657623291,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 2.755859375
        },
        {
          "model": "static",
          "length": 16,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.08499766141176224,
          "query_target_nll": 3.6181375980377197,
          "query_top1_on_support": 0.080078125,
          "mean_query_support_size": 2.611328125,
          "probe_belief_target_mass": 0.0037644123658537865,
          "probe_belief_target_nll": 9.312690734863281,
          "probe_belief_top1_on_support": 0.001953125,
          "mean_belief_support_size": 2.75
        },
        {
          "model": "static",
          "length": 16,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.09336584061384201,
          "query_target_nll": 3.7727713584899902,
          "query_top1_on_support": 0.10546875,
          "mean_query_support_size": 2.79296875,
          "probe_belief_target_mass": 0.00351130124181509,
          "probe_belief_target_nll": 9.531848907470703,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 2.95703125
        },
        {
          "model": "static",
          "length": 24,
          "query_type": "A",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.05222459137439728,
          "query_target_nll": 4.529078483581543,
          "query_top1_on_support": 0.05078125,
          "mean_query_support_size": 1.61328125,
          "probe_belief_target_mass": 0.002090953290462494,
          "probe_belief_target_nll": 8.811631202697754,
          "probe_belief_top1_on_support": 0.001953125,
          "mean_belief_support_size": 1.73828125
        },
        {
          "model": "static",
          "length": 24,
          "query_type": "B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.05711149796843529,
          "query_target_nll": 3.9309914112091064,
          "query_top1_on_support": 0.0546875,
          "mean_query_support_size": 1.654296875,
          "probe_belief_target_mass": 0.002137265168130398,
          "probe_belief_target_nll": 8.818796157836914,
          "probe_belief_top1_on_support": 0.00390625,
          "mean_belief_support_size": 1.73046875
        },
        {
          "model": "static",
          "length": 24,
          "query_type": "A_PLUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.05781635269522667,
          "query_target_nll": 3.5266218185424805,
          "query_top1_on_support": 0.06640625,
          "mean_query_support_size": 1.751953125,
          "probe_belief_target_mass": 0.002095542149618268,
          "probe_belief_target_nll": 8.9423246383667,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 1.8359375
        },
        {
          "model": "static",
          "length": 24,
          "query_type": "A_MINUS_B",
          "k": -1,
          "n": 512,
          "query_target_mass": 0.052211321890354156,
          "query_target_nll": 3.670226573944092,
          "query_top1_on_support": 0.05078125,
          "mean_query_support_size": 1.63671875,
          "probe_belief_target_mass": 0.002095890464261174,
          "probe_belief_target_nll": 9.037616729736328,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 1.72265625
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/dense_latent_query_executor/checkpoints/control_static_mod31/checkpoint_final.pt"
  ]
}