{
  "mode": "dense",
  "variant": "smoke_gru_lowrank_d32_r4",
  "transition": "gru",
  "decoder_type": "low_rank",
  "args": {
    "variant_name": "smoke_gru_lowrank_d32_r4",
    "mode": "dense",
    "supervision": "teacher_belief",
    "seed": 53,
    "modulus": 7,
    "observe_mod": 3,
    "observe_prob": 0.3,
    "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": 32,
    "instr_dim": 32,
    "transition": "gru",
    "decoder_type": "low_rank",
    "decoder_rank": 4,
    "heads": 4,
    "compiler_layers": 2,
    "dropout": 0.0,
    "train_steps": 2,
    "batch_size": 16,
    "eval_batch_size": 16,
    "eval_examples": 16,
    "lr": 0.001,
    "weight_decay": 0.0001,
    "query_loss_weight": 0.0,
    "belief_loss_weight": 1.0,
    "decoder_query_loss_weight": 0.0,
    "belief_distill_prob": 1.0,
    "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_teacher_distillation/runs/smoke_gru_lowrank_d32_r4",
    "checkpoint_dir": "large_artifacts/dense_teacher_distillation/checkpoints/smoke_gru_lowrank_d32_r4"
  },
  "env": {
    "python": "3.12.3",
    "torch": "2.8.0+cu128",
    "device": "cuda",
    "cuda_name": "NVIDIA RTX 6000 Ada Generation"
  },
  "train": [
    {
      "step": 1,
      "loss": 3.907280445098877,
      "grad_norm": 0.31312790513038635,
      "query_loss": 0.0,
      "belief_loss": 3.907280445098877,
      "decoder_query_loss": 0.0,
      "query_target_mass": 0.43061020970344543,
      "query_target_nll": 1.9375801086425781,
      "query_top1_on_support": 0.4375,
      "mean_query_support_size": 3.0,
      "decoder_belief_target_mass": 0.08954167366027832,
      "decoder_belief_target_nll": 3.8769712448120117,
      "decoder_belief_top1_on_support": 0.0625,
      "mean_decoder_belief_support_size": 4.3125,
      "decoder_query_target_mass": 0.42974328994750977,
      "decoder_query_target_nll": 1.9427165985107422,
      "decoder_query_top1_on_support": 0.4375,
      "mean_decoder_query_support_size": 3.0
    },
    {
      "step": 2,
      "loss": 3.8941969871520996,
      "grad_norm": 0.24047841131687164,
      "query_loss": 0.0,
      "belief_loss": 3.8941969871520996,
      "decoder_query_loss": 0.0,
      "query_target_mass": 0.5106407999992371,
      "query_target_nll": 1.9512581825256348,
      "query_top1_on_support": 0.5,
      "mean_query_support_size": 3.5625,
      "decoder_belief_target_mass": 0.09862157702445984,
      "decoder_belief_target_nll": 3.8546619415283203,
      "decoder_belief_top1_on_support": 0.1875,
      "mean_decoder_belief_support_size": 4.6875,
      "decoder_query_target_mass": 0.5147466063499451,
      "decoder_query_target_nll": 1.9293339252471924,
      "decoder_query_top1_on_support": 0.5,
      "mean_decoder_query_support_size": 3.5625
    }
  ],
  "probe_train": [
    {
      "step": 1,
      "loss": 3.9226200580596924,
      "grad_norm": 0.4673333764076233,
      "probe_belief_target_mass": 0.10903073847293854,
      "probe_belief_target_nll": 3.9226200580596924,
      "probe_belief_top1_on_support": 0.125,
      "mean_belief_support_size": 5.359375
    },
    {
      "step": 2,
      "loss": 3.922621965408325,
      "grad_norm": 0.46719542145729065,
      "probe_belief_target_mass": 0.11432923376560211,
      "probe_belief_target_nll": 3.922621965408325,
      "probe_belief_top1_on_support": 0.09375,
      "mean_belief_support_size": 5.65625
    }
  ],
  "eval": [
    {
      "step": 2,
      "rows": [
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.6324458718299866,
          "query_target_nll": 1.9907984733581543,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 4.5,
          "probe_belief_target_mass": 0.09723807871341705,
          "probe_belief_target_nll": 3.857130289077759,
          "probe_belief_top1_on_support": 0.1875,
          "mean_belief_support_size": 4.5,
          "decoder_belief_target_mass": 0.09204012155532837,
          "decoder_belief_target_nll": 3.899122476577759,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.5,
          "decoder_query_target_mass": 0.6472550630569458,
          "decoder_query_target_nll": 1.9347026348114014,
          "decoder_query_top1_on_support": 0.6875,
          "mean_decoder_query_support_size": 4.5
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.6410974264144897,
          "query_target_nll": 1.9620356559753418,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 4.5,
          "probe_belief_target_mass": 0.09498191624879837,
          "probe_belief_target_nll": 3.8751001358032227,
          "probe_belief_top1_on_support": 0.1875,
          "mean_belief_support_size": 4.5,
          "decoder_belief_target_mass": 0.09082555025815964,
          "decoder_belief_target_nll": 3.9172089099884033,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.5,
          "decoder_query_target_mass": 0.6440725326538086,
          "decoder_query_target_nll": 1.943631887435913,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 4.5
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.6391615867614746,
          "query_target_nll": 1.9666028022766113,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 4.5,
          "probe_belief_target_mass": 0.09327226877212524,
          "probe_belief_target_nll": 3.8854923248291016,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.5,
          "decoder_belief_target_mass": 0.09056565910577774,
          "decoder_belief_target_nll": 3.9214048385620117,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.5,
          "decoder_query_target_mass": 0.6428135633468628,
          "decoder_query_target_nll": 1.9488974809646606,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 4.5
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.6391615867614746,
          "query_target_nll": 1.9666028022766113,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 4.5,
          "probe_belief_target_mass": 0.09327226877212524,
          "probe_belief_target_nll": 3.8854923248291016,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.5,
          "decoder_belief_target_mass": 0.09056565910577774,
          "decoder_belief_target_nll": 3.9214048385620117,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.5,
          "decoder_query_target_mass": 0.6428135633468628,
          "decoder_query_target_nll": 1.9488974809646606,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 4.5
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "B",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.60848468542099,
          "query_target_nll": 1.950240969657898,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 4.25,
          "probe_belief_target_mass": 0.09550680220127106,
          "probe_belief_target_nll": 3.923290491104126,
          "probe_belief_top1_on_support": 0.1875,
          "mean_belief_support_size": 4.6875,
          "decoder_belief_target_mass": 0.09354785084724426,
          "decoder_belief_target_nll": 3.9355316162109375,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.6875,
          "decoder_query_target_mass": 0.5971983671188354,
          "decoder_query_target_nll": 1.9892348051071167,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 4.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "B",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.6128774881362915,
          "query_target_nll": 1.9287092685699463,
          "query_top1_on_support": 0.625,
          "mean_query_support_size": 4.25,
          "probe_belief_target_mass": 0.09399012476205826,
          "probe_belief_target_nll": 3.969205379486084,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.6875,
          "decoder_belief_target_mass": 0.0934111475944519,
          "decoder_belief_target_nll": 3.9420714378356934,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.6875,
          "decoder_query_target_mass": 0.596504807472229,
          "decoder_query_target_nll": 1.9921716451644897,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 4.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "B",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.6097608804702759,
          "query_target_nll": 1.9422633647918701,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 4.25,
          "probe_belief_target_mass": 0.09379327297210693,
          "probe_belief_target_nll": 3.985757827758789,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.6875,
          "decoder_belief_target_mass": 0.09480030089616776,
          "decoder_belief_target_nll": 3.9343249797821045,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.6875,
          "decoder_query_target_mass": 0.5971996188163757,
          "decoder_query_target_nll": 1.9879391193389893,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 4.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "B",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.6097608804702759,
          "query_target_nll": 1.9422633647918701,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 4.25,
          "probe_belief_target_mass": 0.09379327297210693,
          "probe_belief_target_nll": 3.985757827758789,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.6875,
          "decoder_belief_target_mass": 0.09480030089616776,
          "decoder_belief_target_nll": 3.9343249797821045,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.6875,
          "decoder_query_target_mass": 0.5971996188163757,
          "decoder_query_target_nll": 1.9879391193389893,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 4.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.6065546274185181,
          "query_target_nll": 1.9876532554626465,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 4.3125,
          "probe_belief_target_mass": 0.08992785215377808,
          "probe_belief_target_nll": 3.9207687377929688,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.3125,
          "decoder_belief_target_mass": 0.08902467042207718,
          "decoder_belief_target_nll": 3.878098487854004,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 4.3125,
          "decoder_query_target_mass": 0.616490364074707,
          "decoder_query_target_nll": 1.9447731971740723,
          "decoder_query_top1_on_support": 0.75,
          "mean_decoder_query_support_size": 4.3125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.6058304309844971,
          "query_target_nll": 1.9998646974563599,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 4.3125,
          "probe_belief_target_mass": 0.0897005945444107,
          "probe_belief_target_nll": 3.9086990356445312,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.3125,
          "decoder_belief_target_mass": 0.0876334011554718,
          "decoder_belief_target_nll": 3.886727809906006,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 4.3125,
          "decoder_query_target_mass": 0.6161534190177917,
          "decoder_query_target_nll": 1.9457006454467773,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 4.3125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.6066025495529175,
          "query_target_nll": 2.0036420822143555,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 4.3125,
          "probe_belief_target_mass": 0.08857643604278564,
          "probe_belief_target_nll": 3.9272546768188477,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.3125,
          "decoder_belief_target_mass": 0.08842548727989197,
          "decoder_belief_target_nll": 3.8848395347595215,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 4.3125,
          "decoder_query_target_mass": 0.6159521341323853,
          "decoder_query_target_nll": 1.9464248418807983,
          "decoder_query_top1_on_support": 0.625,
          "mean_decoder_query_support_size": 4.3125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_PLUS_B",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.6066025495529175,
          "query_target_nll": 2.0036420822143555,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 4.3125,
          "probe_belief_target_mass": 0.08857643604278564,
          "probe_belief_target_nll": 3.9272546768188477,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.3125,
          "decoder_belief_target_mass": 0.08842548727989197,
          "decoder_belief_target_nll": 3.8848395347595215,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 4.3125,
          "decoder_query_target_mass": 0.6159521341323853,
          "decoder_query_target_nll": 1.9464248418807983,
          "decoder_query_top1_on_support": 0.625,
          "mean_decoder_query_support_size": 4.3125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.45503348112106323,
          "query_target_nll": 2.0205912590026855,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 3.25,
          "probe_belief_target_mass": 0.08463456481695175,
          "probe_belief_target_nll": 3.928447723388672,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 4.25,
          "decoder_belief_target_mass": 0.08595141768455505,
          "decoder_belief_target_nll": 3.894428014755249,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.25,
          "decoder_query_target_mass": 0.46443504095077515,
          "decoder_query_target_nll": 1.9449232816696167,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 3.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.4513486921787262,
          "query_target_nll": 2.0389204025268555,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.25,
          "probe_belief_target_mass": 0.08475788682699203,
          "probe_belief_target_nll": 3.9429125785827637,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.25,
          "decoder_belief_target_mass": 0.08580811321735382,
          "decoder_belief_target_nll": 3.9028289318084717,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 4.25,
          "decoder_query_target_mass": 0.4643886685371399,
          "decoder_query_target_nll": 1.9454562664031982,
          "decoder_query_top1_on_support": 0.4375,
          "mean_decoder_query_support_size": 3.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.45170021057128906,
          "query_target_nll": 2.0331814289093018,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 3.25,
          "probe_belief_target_mass": 0.08491122722625732,
          "probe_belief_target_nll": 3.929232120513916,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 4.25,
          "decoder_belief_target_mass": 0.08609213680028915,
          "decoder_belief_target_nll": 3.9069325923919678,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.25,
          "decoder_query_target_mass": 0.46437129378318787,
          "decoder_query_target_nll": 1.9453853368759155,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 3.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 2,
          "query_type": "A_MINUS_B",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.45170021057128906,
          "query_target_nll": 2.0331814289093018,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 3.25,
          "probe_belief_target_mass": 0.08491122722625732,
          "probe_belief_target_nll": 3.929232120513916,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 4.25,
          "decoder_belief_target_mass": 0.08609213680028915,
          "decoder_belief_target_nll": 3.9069325923919678,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.25,
          "decoder_query_target_mass": 0.46437129378318787,
          "decoder_query_target_nll": 1.9453853368759155,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 3.25
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.40985095500946045,
          "query_target_nll": 1.9514131546020508,
          "query_top1_on_support": 0.375,
          "mean_query_support_size": 2.875,
          "probe_belief_target_mass": 0.06511448323726654,
          "probe_belief_target_nll": 3.9094491004943848,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.3125,
          "decoder_belief_target_mass": 0.06773223727941513,
          "decoder_belief_target_nll": 3.9147443771362305,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 3.3125,
          "decoder_query_target_mass": 0.41371044516563416,
          "decoder_query_target_nll": 1.9309883117675781,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 2.875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.4086969494819641,
          "query_target_nll": 1.9681390523910522,
          "query_top1_on_support": 0.3125,
          "mean_query_support_size": 2.875,
          "probe_belief_target_mass": 0.06640087813138962,
          "probe_belief_target_nll": 3.889374017715454,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 3.3125,
          "decoder_belief_target_mass": 0.06847355514764786,
          "decoder_belief_target_nll": 3.889786958694458,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 3.3125,
          "decoder_query_target_mass": 0.4170439839363098,
          "decoder_query_target_nll": 1.9193686246871948,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 2.875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.410377562046051,
          "query_target_nll": 1.960054636001587,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 2.875,
          "probe_belief_target_mass": 0.06528069078922272,
          "probe_belief_target_nll": 3.922804832458496,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 3.3125,
          "decoder_belief_target_mass": 0.0684465616941452,
          "decoder_belief_target_nll": 3.8993027210235596,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 3.3125,
          "decoder_query_target_mass": 0.4139825701713562,
          "decoder_query_target_nll": 1.9326714277267456,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 2.875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.4086545705795288,
          "query_target_nll": 1.967203974723816,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 2.875,
          "probe_belief_target_mass": 0.06647180765867233,
          "probe_belief_target_nll": 3.9012179374694824,
          "probe_belief_top1_on_support": 0.0,
          "mean_belief_support_size": 3.3125,
          "decoder_belief_target_mass": 0.06781641393899918,
          "decoder_belief_target_nll": 3.909724473953247,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 3.3125,
          "decoder_query_target_mass": 0.4120856821537018,
          "decoder_query_target_nll": 1.938500165939331,
          "decoder_query_top1_on_support": 0.4375,
          "mean_decoder_query_support_size": 2.875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "B",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.3985183537006378,
          "query_target_nll": 1.9552505016326904,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 2.75,
          "probe_belief_target_mass": 0.07589889317750931,
          "probe_belief_target_nll": 3.9188568592071533,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.8125,
          "decoder_belief_target_mass": 0.07812336832284927,
          "decoder_belief_target_nll": 3.8838629722595215,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 3.8125,
          "decoder_query_target_mass": 0.4003772735595703,
          "decoder_query_target_nll": 1.9292001724243164,
          "decoder_query_top1_on_support": 0.4375,
          "mean_decoder_query_support_size": 2.75
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "B",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.40383392572402954,
          "query_target_nll": 1.9361836910247803,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 2.75,
          "probe_belief_target_mass": 0.07711724936962128,
          "probe_belief_target_nll": 3.9187417030334473,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 3.8125,
          "decoder_belief_target_mass": 0.07796205580234528,
          "decoder_belief_target_nll": 3.885453462600708,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 3.8125,
          "decoder_query_target_mass": 0.3996056914329529,
          "decoder_query_target_nll": 1.9361112117767334,
          "decoder_query_top1_on_support": 0.4375,
          "mean_decoder_query_support_size": 2.75
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "B",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.404242604970932,
          "query_target_nll": 1.9149739742279053,
          "query_top1_on_support": 0.625,
          "mean_query_support_size": 2.75,
          "probe_belief_target_mass": 0.07924097776412964,
          "probe_belief_target_nll": 3.8776936531066895,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.8125,
          "decoder_belief_target_mass": 0.07955636084079742,
          "decoder_belief_target_nll": 3.8751327991485596,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 3.8125,
          "decoder_query_target_mass": 0.3998994529247284,
          "decoder_query_target_nll": 1.9386790990829468,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 2.75
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "B",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.4014626145362854,
          "query_target_nll": 1.9203784465789795,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 2.75,
          "probe_belief_target_mass": 0.07771638035774231,
          "probe_belief_target_nll": 3.8957011699676514,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.8125,
          "decoder_belief_target_mass": 0.0771244615316391,
          "decoder_belief_target_nll": 3.8999462127685547,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 3.8125,
          "decoder_query_target_mass": 0.39769497513771057,
          "decoder_query_target_nll": 1.9508333206176758,
          "decoder_query_top1_on_support": 0.4375,
          "mean_decoder_query_support_size": 2.75
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.4622079133987427,
          "query_target_nll": 1.948323130607605,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.06370759010314941,
          "probe_belief_target_nll": 3.9055962562561035,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.1875,
          "decoder_belief_target_mass": 0.06411153078079224,
          "decoder_belief_target_nll": 3.9109442234039307,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 3.1875,
          "decoder_query_target_mass": 0.45569151639938354,
          "decoder_query_target_nll": 1.9445571899414062,
          "decoder_query_top1_on_support": 0.6875,
          "mean_decoder_query_support_size": 3.1875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.45845192670822144,
          "query_target_nll": 1.9593706130981445,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.06361007690429688,
          "probe_belief_target_nll": 3.8995985984802246,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.1875,
          "decoder_belief_target_mass": 0.06419327110052109,
          "decoder_belief_target_nll": 3.921056032180786,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 3.1875,
          "decoder_query_target_mass": 0.45524996519088745,
          "decoder_query_target_nll": 1.9457900524139404,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 3.1875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.4542197287082672,
          "query_target_nll": 1.9778485298156738,
          "query_top1_on_support": 0.4375,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.06382286548614502,
          "probe_belief_target_nll": 3.8989837169647217,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 3.1875,
          "decoder_belief_target_mass": 0.06473660469055176,
          "decoder_belief_target_nll": 3.9145395755767822,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 3.1875,
          "decoder_query_target_mass": 0.45584404468536377,
          "decoder_query_target_nll": 1.9437503814697266,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 3.1875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_PLUS_B",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.45421838760375977,
          "query_target_nll": 1.9845020771026611,
          "query_top1_on_support": 0.5,
          "mean_query_support_size": 3.1875,
          "probe_belief_target_mass": 0.0638408362865448,
          "probe_belief_target_nll": 3.913921594619751,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 3.1875,
          "decoder_belief_target_mass": 0.06494272500276566,
          "decoder_belief_target_nll": 3.9081003665924072,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 3.1875,
          "decoder_query_target_mass": 0.45552322268486023,
          "decoder_query_target_nll": 1.9449169635772705,
          "decoder_query_top1_on_support": 0.5,
          "mean_decoder_query_support_size": 3.1875
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 0,
          "n": 16,
          "query_target_mass": 0.541875958442688,
          "query_target_nll": 1.9599800109863281,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 3.8125,
          "probe_belief_target_mass": 0.09275917708873749,
          "probe_belief_target_nll": 3.9302711486816406,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.625,
          "decoder_belief_target_mass": 0.09320418536663055,
          "decoder_belief_target_nll": 3.9090871810913086,
          "decoder_belief_top1_on_support": 0.0625,
          "mean_decoder_belief_support_size": 4.625,
          "decoder_query_target_mass": 0.5446874499320984,
          "decoder_query_target_nll": 1.9454948902130127,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 3.8125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 1,
          "n": 16,
          "query_target_mass": 0.5462799072265625,
          "query_target_nll": 1.9471790790557861,
          "query_top1_on_support": 0.5625,
          "mean_query_support_size": 3.8125,
          "probe_belief_target_mass": 0.09457217901945114,
          "probe_belief_target_nll": 3.9167420864105225,
          "probe_belief_top1_on_support": 0.125,
          "mean_belief_support_size": 4.625,
          "decoder_belief_target_mass": 0.09211918711662292,
          "decoder_belief_target_nll": 3.917644739151001,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.625,
          "decoder_query_target_mass": 0.5445622801780701,
          "decoder_query_target_nll": 1.9466599225997925,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 3.8125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 2,
          "n": 16,
          "query_target_mass": 0.5456491708755493,
          "query_target_nll": 1.948580026626587,
          "query_top1_on_support": 0.625,
          "mean_query_support_size": 3.8125,
          "probe_belief_target_mass": 0.09399498999118805,
          "probe_belief_target_nll": 3.9231090545654297,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.625,
          "decoder_belief_target_mass": 0.09380091726779938,
          "decoder_belief_target_nll": 3.9028353691101074,
          "decoder_belief_top1_on_support": 0.0,
          "mean_decoder_belief_support_size": 4.625,
          "decoder_query_target_mass": 0.5444324016571045,
          "decoder_query_target_nll": 1.947416067123413,
          "decoder_query_top1_on_support": 0.5625,
          "mean_decoder_query_support_size": 3.8125
        },
        {
          "model": "dense",
          "variant": "smoke_gru_lowrank_d32_r4",
          "supervision": "teacher_belief",
          "transition": "gru",
          "decoder_type": "low_rank",
          "state_dim": 32,
          "instr_dim": 32,
          "decoder_rank": 4,
          "length": 3,
          "query_type": "A_MINUS_B",
          "k": 3,
          "n": 16,
          "query_target_mass": 0.549125075340271,
          "query_target_nll": 1.9436267614364624,
          "query_top1_on_support": 0.625,
          "mean_query_support_size": 3.8125,
          "probe_belief_target_mass": 0.09527835249900818,
          "probe_belief_target_nll": 3.9071297645568848,
          "probe_belief_top1_on_support": 0.0625,
          "mean_belief_support_size": 4.625,
          "decoder_belief_target_mass": 0.09423324465751648,
          "decoder_belief_target_nll": 3.8979909420013428,
          "decoder_belief_top1_on_support": 0.125,
          "mean_decoder_belief_support_size": 4.625,
          "decoder_query_target_mass": 0.5442381501197815,
          "decoder_query_target_nll": 1.9482319355010986,
          "decoder_query_top1_on_support": 0.4375,
          "mean_decoder_query_support_size": 3.8125
        }
      ]
    }
  ],
  "checkpoints": [
    "large_artifacts/dense_teacher_distillation/checkpoints/smoke_gru_lowrank_d32_r4/checkpoint_final.pt"
  ]
}