{ "mode": "dense", "variant": "smoke_residual_mlp_d32", "transition": "residual", "decoder_type": "mlp", "args": { "variant_name": "smoke_residual_mlp_d32", "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": "residual", "decoder_type": "mlp", "decoder_rank": 16, "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_residual_mlp_d32", "checkpoint_dir": "large_artifacts/dense_teacher_distillation/checkpoints/smoke_residual_mlp_d32" }, "env": { "python": "3.12.3", "torch": "2.8.0+cu128", "device": "cuda", "cuda_name": "NVIDIA RTX 6000 Ada Generation" }, "train": [ { "step": 1, "loss": 3.9155490398406982, "grad_norm": 0.478115439414978, "query_loss": 0.0, "belief_loss": 3.9155490398406982, "decoder_query_loss": 0.0, "query_target_mass": 0.41651859879493713, "query_target_nll": 2.019354820251465, "query_top1_on_support": 0.375, "mean_query_support_size": 3.0, "decoder_belief_target_mass": 0.0929194837808609, "decoder_belief_target_nll": 3.860851287841797, "decoder_belief_top1_on_support": 0.125, "mean_decoder_belief_support_size": 4.3125, "decoder_query_target_mass": 0.4338352680206299, "decoder_query_target_nll": 1.9293948411941528, "decoder_query_top1_on_support": 0.4375, "mean_decoder_query_support_size": 3.0 }, { "step": 2, "loss": 3.8746559619903564, "grad_norm": 0.47203755378723145, "query_loss": 0.0, "belief_loss": 3.8746559619903564, "decoder_query_loss": 0.0, "query_target_mass": 0.52340167760849, "query_target_nll": 1.8962289094924927, "query_top1_on_support": 0.625, "mean_query_support_size": 3.5625, "decoder_belief_target_mass": 0.09804157167673111, "decoder_belief_target_nll": 3.858799934387207, "decoder_belief_top1_on_support": 0.1875, "mean_decoder_belief_support_size": 4.6875, "decoder_query_target_mass": 0.5091164708137512, "decoder_query_target_nll": 1.9527616500854492, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 3.5625 } ], "probe_train": [ { "step": 1, "loss": 3.8708853721618652, "grad_norm": 0.4521472454071045, "probe_belief_target_mass": 0.11238079518079758, "probe_belief_target_nll": 3.8708853721618652, "probe_belief_top1_on_support": 0.15625, "mean_belief_support_size": 5.359375 }, { "step": 2, "loss": 3.8747031688690186, "grad_norm": 0.4354128837585449, "probe_belief_target_mass": 0.11844587326049805, "probe_belief_target_nll": 3.8747031688690186, "probe_belief_top1_on_support": 0.15625, "mean_belief_support_size": 5.65625 } ], "eval": [ { "step": 2, "rows": [ { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A", "k": 0, "n": 16, "query_target_mass": 0.635539174079895, "query_target_nll": 1.987573266029358, "query_top1_on_support": 0.5625, "mean_query_support_size": 4.5, "probe_belief_target_mass": 0.09267168492078781, "probe_belief_target_nll": 3.904350757598877, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.5, "decoder_belief_target_mass": 0.09193269908428192, "decoder_belief_target_nll": 3.9068260192871094, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.5, "decoder_query_target_mass": 0.6445528268814087, "decoder_query_target_nll": 1.9418132305145264, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.5 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A", "k": 1, "n": 16, "query_target_mass": 0.6373782157897949, "query_target_nll": 1.9818055629730225, "query_top1_on_support": 0.5625, "mean_query_support_size": 4.5, "probe_belief_target_mass": 0.09264657646417618, "probe_belief_target_nll": 3.904862403869629, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.5, "decoder_belief_target_mass": 0.09220969676971436, "decoder_belief_target_nll": 3.9009432792663574, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.5, "decoder_query_target_mass": 0.64577716588974, "decoder_query_target_nll": 1.9376941919326782, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.5 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A", "k": 2, "n": 16, "query_target_mass": 0.6365286111831665, "query_target_nll": 1.986505389213562, "query_top1_on_support": 0.6875, "mean_query_support_size": 4.5, "probe_belief_target_mass": 0.09282328188419342, "probe_belief_target_nll": 3.904597759246826, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.5, "decoder_belief_target_mass": 0.09237661957740784, "decoder_belief_target_nll": 3.8977274894714355, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.5, "decoder_query_target_mass": 0.6455986499786377, "decoder_query_target_nll": 1.9383410215377808, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.5 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A", "k": 3, "n": 16, "query_target_mass": 0.6365286111831665, "query_target_nll": 1.986505389213562, "query_top1_on_support": 0.6875, "mean_query_support_size": 4.5, "probe_belief_target_mass": 0.09282328188419342, "probe_belief_target_nll": 3.904597759246826, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.5, "decoder_belief_target_mass": 0.09237661957740784, "decoder_belief_target_nll": 3.8977274894714355, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.5, "decoder_query_target_mass": 0.6455986499786377, "decoder_query_target_nll": 1.9383410215377808, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.5 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "B", "k": 0, "n": 16, "query_target_mass": 0.6154197454452515, "query_target_nll": 1.9321234226226807, "query_top1_on_support": 0.5625, "mean_query_support_size": 4.25, "probe_belief_target_mass": 0.09975500404834747, "probe_belief_target_nll": 3.8781609535217285, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.6875, "decoder_belief_target_mass": 0.09665153920650482, "decoder_belief_target_nll": 3.9237911701202393, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.6875, "decoder_query_target_mass": 0.6043015718460083, "decoder_query_target_nll": 1.9576983451843262, "decoder_query_top1_on_support": 0.5, "mean_decoder_query_support_size": 4.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "B", "k": 1, "n": 16, "query_target_mass": 0.6140295267105103, "query_target_nll": 1.9351900815963745, "query_top1_on_support": 0.5, "mean_query_support_size": 4.25, "probe_belief_target_mass": 0.0995228961110115, "probe_belief_target_nll": 3.8774921894073486, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.6875, "decoder_belief_target_mass": 0.09670792520046234, "decoder_belief_target_nll": 3.9232382774353027, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.6875, "decoder_query_target_mass": 0.6043061017990112, "decoder_query_target_nll": 1.9579203128814697, "decoder_query_top1_on_support": 0.5625, "mean_decoder_query_support_size": 4.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "B", "k": 2, "n": 16, "query_target_mass": 0.6136249303817749, "query_target_nll": 1.936144232749939, "query_top1_on_support": 0.5625, "mean_query_support_size": 4.25, "probe_belief_target_mass": 0.09950849413871765, "probe_belief_target_nll": 3.877800226211548, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.6875, "decoder_belief_target_mass": 0.09640902280807495, "decoder_belief_target_nll": 3.9272027015686035, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.6875, "decoder_query_target_mass": 0.6045119166374207, "decoder_query_target_nll": 1.9571914672851562, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "B", "k": 3, "n": 16, "query_target_mass": 0.6136249303817749, "query_target_nll": 1.936144232749939, "query_top1_on_support": 0.5625, "mean_query_support_size": 4.25, "probe_belief_target_mass": 0.09950849413871765, "probe_belief_target_nll": 3.877800226211548, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.6875, "decoder_belief_target_mass": 0.09640902280807495, "decoder_belief_target_nll": 3.9272027015686035, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.6875, "decoder_query_target_mass": 0.6045119166374207, "decoder_query_target_nll": 1.9571914672851562, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_PLUS_B", "k": 0, "n": 16, "query_target_mass": 0.6133795380592346, "query_target_nll": 1.9749605655670166, "query_top1_on_support": 0.5625, "mean_query_support_size": 4.3125, "probe_belief_target_mass": 0.09025685489177704, "probe_belief_target_nll": 3.8758795261383057, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.3125, "decoder_belief_target_mass": 0.08517290651798248, "decoder_belief_target_nll": 3.943145751953125, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.3125, "decoder_query_target_mass": 0.6173381805419922, "decoder_query_target_nll": 1.9440617561340332, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.3125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_PLUS_B", "k": 1, "n": 16, "query_target_mass": 0.6122944951057434, "query_target_nll": 1.9789464473724365, "query_top1_on_support": 0.625, "mean_query_support_size": 4.3125, "probe_belief_target_mass": 0.09059484302997589, "probe_belief_target_nll": 3.872110366821289, "probe_belief_top1_on_support": 0.125, "mean_belief_support_size": 4.3125, "decoder_belief_target_mass": 0.08575474470853806, "decoder_belief_target_nll": 3.935415029525757, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.3125, "decoder_query_target_mass": 0.6168403625488281, "decoder_query_target_nll": 1.945664882659912, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.3125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_PLUS_B", "k": 2, "n": 16, "query_target_mass": 0.6118228435516357, "query_target_nll": 1.981611728668213, "query_top1_on_support": 0.625, "mean_query_support_size": 4.3125, "probe_belief_target_mass": 0.09070192277431488, "probe_belief_target_nll": 3.8711366653442383, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.3125, "decoder_belief_target_mass": 0.08611176162958145, "decoder_belief_target_nll": 3.9303629398345947, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.3125, "decoder_query_target_mass": 0.6166759729385376, "decoder_query_target_nll": 1.9464406967163086, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.3125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_PLUS_B", "k": 3, "n": 16, "query_target_mass": 0.6118228435516357, "query_target_nll": 1.981611728668213, "query_top1_on_support": 0.625, "mean_query_support_size": 4.3125, "probe_belief_target_mass": 0.09070192277431488, "probe_belief_target_nll": 3.8711366653442383, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.3125, "decoder_belief_target_mass": 0.08611176162958145, "decoder_belief_target_nll": 3.9303629398345947, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.3125, "decoder_query_target_mass": 0.6166759729385376, "decoder_query_target_nll": 1.9464406967163086, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 4.3125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_MINUS_B", "k": 0, "n": 16, "query_target_mass": 0.4616420269012451, "query_target_nll": 1.9784964323043823, "query_top1_on_support": 0.4375, "mean_query_support_size": 3.25, "probe_belief_target_mass": 0.0875224620103836, "probe_belief_target_nll": 3.9266743659973145, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.25, "decoder_belief_target_mass": 0.08843902498483658, "decoder_belief_target_nll": 3.896263599395752, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.25, "decoder_query_target_mass": 0.46323490142822266, "decoder_query_target_nll": 1.9481515884399414, "decoder_query_top1_on_support": 0.4375, "mean_decoder_query_support_size": 3.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_MINUS_B", "k": 1, "n": 16, "query_target_mass": 0.464134156703949, "query_target_nll": 1.9630892276763916, "query_top1_on_support": 0.4375, "mean_query_support_size": 3.25, "probe_belief_target_mass": 0.08819176256656647, "probe_belief_target_nll": 3.9215359687805176, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.25, "decoder_belief_target_mass": 0.08877000212669373, "decoder_belief_target_nll": 3.895047903060913, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.25, "decoder_query_target_mass": 0.463310569524765, "decoder_query_target_nll": 1.9478578567504883, "decoder_query_top1_on_support": 0.375, "mean_decoder_query_support_size": 3.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_MINUS_B", "k": 2, "n": 16, "query_target_mass": 0.46407920122146606, "query_target_nll": 1.9640332460403442, "query_top1_on_support": 0.4375, "mean_query_support_size": 3.25, "probe_belief_target_mass": 0.08824597299098969, "probe_belief_target_nll": 3.9222686290740967, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.25, "decoder_belief_target_mass": 0.08895158767700195, "decoder_belief_target_nll": 3.893507242202759, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.25, "decoder_query_target_mass": 0.4633744955062866, "decoder_query_target_nll": 1.9477055072784424, "decoder_query_top1_on_support": 0.4375, "mean_decoder_query_support_size": 3.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 2, "query_type": "A_MINUS_B", "k": 3, "n": 16, "query_target_mass": 0.46407920122146606, "query_target_nll": 1.9640332460403442, "query_top1_on_support": 0.4375, "mean_query_support_size": 3.25, "probe_belief_target_mass": 0.08824597299098969, "probe_belief_target_nll": 3.9222686290740967, "probe_belief_top1_on_support": 0.1875, "mean_belief_support_size": 4.25, "decoder_belief_target_mass": 0.08895158767700195, "decoder_belief_target_nll": 3.893507242202759, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.25, "decoder_query_target_mass": 0.46337446570396423, "decoder_query_target_nll": 1.9477055072784424, "decoder_query_top1_on_support": 0.4375, "mean_decoder_query_support_size": 3.25 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A", "k": 0, "n": 16, "query_target_mass": 0.4078540802001953, "query_target_nll": 1.966007947921753, "query_top1_on_support": 0.4375, "mean_query_support_size": 2.875, "probe_belief_target_mass": 0.06950856745243073, "probe_belief_target_nll": 3.885192632675171, "probe_belief_top1_on_support": 0.125, "mean_belief_support_size": 3.3125, "decoder_belief_target_mass": 0.06611587852239609, "decoder_belief_target_nll": 3.9092555046081543, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.3125, "decoder_query_target_mass": 0.41071856021881104, "decoder_query_target_nll": 1.94843590259552, "decoder_query_top1_on_support": 0.375, "mean_decoder_query_support_size": 2.875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A", "k": 1, "n": 16, "query_target_mass": 0.4093289375305176, "query_target_nll": 1.9602549076080322, "query_top1_on_support": 0.375, "mean_query_support_size": 2.875, "probe_belief_target_mass": 0.06983628869056702, "probe_belief_target_nll": 3.883185863494873, "probe_belief_top1_on_support": 0.125, "mean_belief_support_size": 3.3125, "decoder_belief_target_mass": 0.06623448431491852, "decoder_belief_target_nll": 3.90468692779541, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.3125, "decoder_query_target_mass": 0.4110081195831299, "decoder_query_target_nll": 1.9481414556503296, "decoder_query_top1_on_support": 0.375, "mean_decoder_query_support_size": 2.875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A", "k": 2, "n": 16, "query_target_mass": 0.41037529706954956, "query_target_nll": 1.9572917222976685, "query_top1_on_support": 0.375, "mean_query_support_size": 2.875, "probe_belief_target_mass": 0.07003416121006012, "probe_belief_target_nll": 3.8818745613098145, "probe_belief_top1_on_support": 0.125, "mean_belief_support_size": 3.3125, "decoder_belief_target_mass": 0.06643395870923996, "decoder_belief_target_nll": 3.902160406112671, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.3125, "decoder_query_target_mass": 0.41104578971862793, "decoder_query_target_nll": 1.9486020803451538, "decoder_query_top1_on_support": 0.3125, "mean_decoder_query_support_size": 2.875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A", "k": 3, "n": 16, "query_target_mass": 0.41068899631500244, "query_target_nll": 1.9561498165130615, "query_top1_on_support": 0.3125, "mean_query_support_size": 2.875, "probe_belief_target_mass": 0.07036568224430084, "probe_belief_target_nll": 3.878657341003418, "probe_belief_top1_on_support": 0.125, "mean_belief_support_size": 3.3125, "decoder_belief_target_mass": 0.06652108579874039, "decoder_belief_target_nll": 3.8989949226379395, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.3125, "decoder_query_target_mass": 0.41137486696243286, "decoder_query_target_nll": 1.9480259418487549, "decoder_query_top1_on_support": 0.3125, "mean_decoder_query_support_size": 2.875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "B", "k": 0, "n": 16, "query_target_mass": 0.38965457677841187, "query_target_nll": 1.9609581232070923, "query_top1_on_support": 0.4375, "mean_query_support_size": 2.75, "probe_belief_target_mass": 0.07836291193962097, "probe_belief_target_nll": 3.876579761505127, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.8125, "decoder_belief_target_mass": 0.07649758458137512, "decoder_belief_target_nll": 3.9225568771362305, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.8125, "decoder_query_target_mass": 0.39189213514328003, "decoder_query_target_nll": 1.954052209854126, "decoder_query_top1_on_support": 0.375, "mean_decoder_query_support_size": 2.75 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "B", "k": 1, "n": 16, "query_target_mass": 0.38969331979751587, "query_target_nll": 1.959975242614746, "query_top1_on_support": 0.4375, "mean_query_support_size": 2.75, "probe_belief_target_mass": 0.07811713218688965, "probe_belief_target_nll": 3.877216339111328, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.8125, "decoder_belief_target_mass": 0.07631310820579529, "decoder_belief_target_nll": 3.926846504211426, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.8125, "decoder_query_target_mass": 0.39172178506851196, "decoder_query_target_nll": 1.9552240371704102, "decoder_query_top1_on_support": 0.3125, "mean_decoder_query_support_size": 2.75 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "B", "k": 2, "n": 16, "query_target_mass": 0.38953256607055664, "query_target_nll": 1.9613488912582397, "query_top1_on_support": 0.375, "mean_query_support_size": 2.75, "probe_belief_target_mass": 0.07825909554958344, "probe_belief_target_nll": 3.875828266143799, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.8125, "decoder_belief_target_mass": 0.07654145359992981, "decoder_belief_target_nll": 3.9206337928771973, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.8125, "decoder_query_target_mass": 0.39203333854675293, "decoder_query_target_nll": 1.9538120031356812, "decoder_query_top1_on_support": 0.375, "mean_decoder_query_support_size": 2.75 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "B", "k": 3, "n": 16, "query_target_mass": 0.390896737575531, "query_target_nll": 1.9555370807647705, "query_top1_on_support": 0.375, "mean_query_support_size": 2.75, "probe_belief_target_mass": 0.0782938301563263, "probe_belief_target_nll": 3.87669038772583, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.8125, "decoder_belief_target_mass": 0.07684379816055298, "decoder_belief_target_nll": 3.9155614376068115, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.8125, "decoder_query_target_mass": 0.3921835124492645, "decoder_query_target_nll": 1.9539828300476074, "decoder_query_top1_on_support": 0.375, "mean_decoder_query_support_size": 2.75 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_PLUS_B", "k": 0, "n": 16, "query_target_mass": 0.4481062889099121, "query_target_nll": 2.017754554748535, "query_top1_on_support": 0.375, "mean_query_support_size": 3.1875, "probe_belief_target_mass": 0.06562228500843048, "probe_belief_target_nll": 3.882507801055908, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.1875, "decoder_belief_target_mass": 0.06566132605075836, "decoder_belief_target_nll": 3.8881051540374756, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 3.1875, "decoder_query_target_mass": 0.45680418610572815, "decoder_query_target_nll": 1.9461383819580078, "decoder_query_top1_on_support": 0.5, "mean_decoder_query_support_size": 3.1875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_PLUS_B", "k": 1, "n": 16, "query_target_mass": 0.44766491651535034, "query_target_nll": 2.0158355236053467, "query_top1_on_support": 0.375, "mean_query_support_size": 3.1875, "probe_belief_target_mass": 0.0656525045633316, "probe_belief_target_nll": 3.8878567218780518, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.1875, "decoder_belief_target_mass": 0.06517162173986435, "decoder_belief_target_nll": 3.890270471572876, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 3.1875, "decoder_query_target_mass": 0.4559805393218994, "decoder_query_target_nll": 1.950345754623413, "decoder_query_top1_on_support": 0.5625, "mean_decoder_query_support_size": 3.1875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_PLUS_B", "k": 2, "n": 16, "query_target_mass": 0.4473203420639038, "query_target_nll": 2.0161566734313965, "query_top1_on_support": 0.375, "mean_query_support_size": 3.1875, "probe_belief_target_mass": 0.06567637622356415, "probe_belief_target_nll": 3.8897032737731934, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.1875, "decoder_belief_target_mass": 0.06519179046154022, "decoder_belief_target_nll": 3.890232563018799, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 3.1875, "decoder_query_target_mass": 0.45550307631492615, "decoder_query_target_nll": 1.9528086185455322, "decoder_query_top1_on_support": 0.5625, "mean_decoder_query_support_size": 3.1875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_PLUS_B", "k": 3, "n": 16, "query_target_mass": 0.44661745429039, "query_target_nll": 2.016641139984131, "query_top1_on_support": 0.4375, "mean_query_support_size": 3.1875, "probe_belief_target_mass": 0.06532050669193268, "probe_belief_target_nll": 3.8965024948120117, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 3.1875, "decoder_belief_target_mass": 0.06483437865972519, "decoder_belief_target_nll": 3.8958563804626465, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 3.1875, "decoder_query_target_mass": 0.4553671181201935, "decoder_query_target_nll": 1.953223466873169, "decoder_query_top1_on_support": 0.5, "mean_decoder_query_support_size": 3.1875 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_MINUS_B", "k": 0, "n": 16, "query_target_mass": 0.5365866422653198, "query_target_nll": 1.995911955833435, "query_top1_on_support": 0.5, "mean_query_support_size": 3.8125, "probe_belief_target_mass": 0.09407086670398712, "probe_belief_target_nll": 3.912867546081543, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.625, "decoder_belief_target_mass": 0.0961739793419838, "decoder_belief_target_nll": 3.874502182006836, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.625, "decoder_query_target_mass": 0.5487311482429504, "decoder_query_target_nll": 1.9341444969177246, "decoder_query_top1_on_support": 0.5625, "mean_decoder_query_support_size": 3.8125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_MINUS_B", "k": 1, "n": 16, "query_target_mass": 0.5372793078422546, "query_target_nll": 1.9935036897659302, "query_top1_on_support": 0.5, "mean_query_support_size": 3.8125, "probe_belief_target_mass": 0.09347034990787506, "probe_belief_target_nll": 3.9196648597717285, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.625, "decoder_belief_target_mass": 0.09643887728452682, "decoder_belief_target_nll": 3.872685194015503, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.625, "decoder_query_target_mass": 0.5491207838058472, "decoder_query_target_nll": 1.9325730800628662, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 3.8125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_MINUS_B", "k": 2, "n": 16, "query_target_mass": 0.5372024178504944, "query_target_nll": 1.994643211364746, "query_top1_on_support": 0.5, "mean_query_support_size": 3.8125, "probe_belief_target_mass": 0.09348494559526443, "probe_belief_target_nll": 3.92090106010437, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.625, "decoder_belief_target_mass": 0.09650737792253494, "decoder_belief_target_nll": 3.87226939201355, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 4.625, "decoder_query_target_mass": 0.5491734743118286, "decoder_query_target_nll": 1.9322819709777832, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 3.8125 }, { "model": "dense", "variant": "smoke_residual_mlp_d32", "supervision": "teacher_belief", "transition": "residual", "decoder_type": "mlp", "state_dim": 32, "instr_dim": 32, "decoder_rank": 16, "length": 3, "query_type": "A_MINUS_B", "k": 3, "n": 16, "query_target_mass": 0.5376810431480408, "query_target_nll": 1.991788625717163, "query_top1_on_support": 0.5, "mean_query_support_size": 3.8125, "probe_belief_target_mass": 0.09326629340648651, "probe_belief_target_nll": 3.9240550994873047, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 4.625, "decoder_belief_target_mass": 0.09652358293533325, "decoder_belief_target_nll": 3.871918201446533, "decoder_belief_top1_on_support": 0.0, "mean_decoder_belief_support_size": 4.625, "decoder_query_target_mass": 0.5490949153900146, "decoder_query_target_nll": 1.9324123859405518, "decoder_query_top1_on_support": 0.625, "mean_decoder_query_support_size": 3.8125 } ] } ], "checkpoints": [ "large_artifacts/dense_teacher_distillation/checkpoints/smoke_residual_mlp_d32/checkpoint_final.pt" ] }