{ "mode": "dense", "args": { "mode": "dense", "supervision": "sparse_belief", "seed": 53, "modulus": 11, "observe_mod": 4, "observe_prob": 0.3, "op_family": "full", "query_family": "all", "train_min_len": 1, "train_max_len": 6, "eval_lengths": "3,6,9,12", "eval_k": "0,1,2,3,6,9,12", "eval_query_types": "all", "state_dim": 256, "instr_dim": 128, "heads": 4, "compiler_layers": 2, "dropout": 0.0, "train_steps": 1200, "batch_size": 256, "eval_batch_size": 256, "eval_examples": 256, "lr": 0.001, "weight_decay": 0.0001, "query_loss_weight": 1.0, "belief_loss_weight": 1.0, "belief_distill_prob": 0.1, "probe_steps": 600, "probe_batch_size": 256, "probe_lr": 0.001, "probe_weight_decay": 0.0001, "probe_dropout": 0.0, "max_grad_norm": 1.0, "log_every": 200, "probe_log_every": 100, "output_dir": "experiments/dense_supervision_ladder/runs/pilot_sparse_belief_mod11", "checkpoint_dir": "large_artifacts/dense_supervision_ladder/checkpoints/pilot_sparse_belief_mod11" }, "env": { "python": "3.12.3", "torch": "2.8.0+cu128", "device": "cuda", "cuda_name": "NVIDIA RTX 6000 Ada Generation" }, "train": [ { "step": 1, "loss": 7.22113037109375, "grad_norm": 0.7609311938285828, "query_loss": 2.413424491882324, "belief_loss": 4.807705879211426, "query_target_mass": 0.436489075422287, "query_target_nll": 2.3261332511901855, "query_top1_on_support": 0.55078125, "mean_query_support_size": 4.65625 }, { "step": 200, "loss": 5.313043594360352, "grad_norm": 0.8590790629386902, "query_loss": 2.0478515625, "belief_loss": 3.2651922702789307, "query_target_mass": 0.6038777232170105, "query_target_nll": 2.077441930770874, "query_top1_on_support": 0.640625, "mean_query_support_size": 5.20703125 }, { "step": 400, "loss": 4.68193244934082, "grad_norm": 1.388948678970337, "query_loss": 1.8297314643859863, "belief_loss": 2.852200984954834, "query_target_mass": 0.6186004281044006, "query_target_nll": 1.9085142612457275, "query_top1_on_support": 0.6953125, "mean_query_support_size": 4.734375 }, { "step": 600, "loss": 4.006976127624512, "grad_norm": 1.3269375562667847, "query_loss": 1.7050237655639648, "belief_loss": 2.3019521236419678, "query_target_mass": 0.7666647434234619, "query_target_nll": 1.6418390274047852, "query_top1_on_support": 0.84765625, "mean_query_support_size": 4.85546875 }, { "step": 800, "loss": 3.824230909347534, "grad_norm": 1.1859867572784424, "query_loss": 1.6285510063171387, "belief_loss": 2.1956799030303955, "query_target_mass": 0.8091198205947876, "query_target_nll": 1.5836857557296753, "query_top1_on_support": 0.8828125, "mean_query_support_size": 5.1171875 }, { "step": 1000, "loss": 3.7616066932678223, "grad_norm": 0.9813265800476074, "query_loss": 1.558565378189087, "belief_loss": 2.2030413150787354, "query_target_mass": 0.8009382486343384, "query_target_nll": 1.5421077013015747, "query_top1_on_support": 0.8828125, "mean_query_support_size": 4.8125 }, { "step": 1200, "loss": 3.5850119590759277, "grad_norm": 0.9375258088111877, "query_loss": 1.5453680753707886, "belief_loss": 2.0396437644958496, "query_target_mass": 0.8266642689704895, "query_target_nll": 1.4455652236938477, "query_top1_on_support": 0.8984375, "mean_query_support_size": 4.62890625 } ], "probe_train": [ { "step": 1, "loss": 4.812735080718994, "grad_norm": 0.5407384634017944, "probe_belief_target_mass": 0.06026485562324524, "probe_belief_target_nll": 4.812735080718994, "probe_belief_top1_on_support": 0.0440848246216774, "mean_belief_support_size": 7.3058037757873535 }, { "step": 100, "loss": 2.3084843158721924, "grad_norm": 0.33394405245780945, "probe_belief_target_mass": 0.7745964527130127, "probe_belief_target_nll": 2.3084843158721924, "probe_belief_top1_on_support": 0.8448660969734192, "mean_belief_support_size": 7.245535850524902 }, { "step": 200, "loss": 2.215435028076172, "grad_norm": 0.32267022132873535, "probe_belief_target_mass": 0.8216588497161865, "probe_belief_target_nll": 2.215435028076172, "probe_belief_top1_on_support": 0.8694196939468384, "mean_belief_support_size": 7.470424652099609 }, { "step": 300, "loss": 2.1113338470458984, "grad_norm": 0.31187739968299866, "probe_belief_target_mass": 0.8532765507698059, "probe_belief_target_nll": 2.1113338470458984, "probe_belief_top1_on_support": 0.8978794813156128, "mean_belief_support_size": 7.313058376312256 }, { "step": 400, "loss": 2.130629301071167, "grad_norm": 0.32416433095932007, "probe_belief_target_mass": 0.8301875591278076, "probe_belief_target_nll": 2.130629301071167, "probe_belief_top1_on_support": 0.8861607313156128, "mean_belief_support_size": 7.136160850524902 }, { "step": 500, "loss": 2.0202836990356445, "grad_norm": 0.3032795190811157, "probe_belief_target_mass": 0.8684308528900146, "probe_belief_target_nll": 2.0202836990356445, "probe_belief_top1_on_support": 0.9246652126312256, "mean_belief_support_size": 7.076451301574707 }, { "step": 600, "loss": 2.0993738174438477, "grad_norm": 0.2957102954387665, "probe_belief_target_mass": 0.858248770236969, "probe_belief_target_nll": 2.0993738174438477, "probe_belief_top1_on_support": 0.9079241752624512, "mean_belief_support_size": 7.327567100524902 } ], "eval": [ { "step": 1200, "rows": [ { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 0, "n": 256, "query_target_mass": 0.4712447226047516, "query_target_nll": 2.4046144485473633, "query_top1_on_support": 0.46875, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.06375250220298767, "probe_belief_target_nll": 14.540451049804688, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.06311692297458649, "decoder_belief_target_nll": 14.436793327331543, "decoder_belief_top1_on_support": 0.05078125, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 1, "n": 256, "query_target_mass": 0.5559136271476746, "query_target_nll": 3.595909595489502, "query_top1_on_support": 0.5703125, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.12088948488235474, "probe_belief_target_nll": 12.3599853515625, "probe_belief_top1_on_support": 0.11328125, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.12184295803308487, "decoder_belief_target_nll": 11.18724250793457, "decoder_belief_top1_on_support": 0.12890625, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 2, "n": 256, "query_target_mass": 0.6834572553634644, "query_target_nll": 2.939244270324707, "query_top1_on_support": 0.70703125, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.2515850365161896, "probe_belief_target_nll": 8.249540328979492, "probe_belief_top1_on_support": 0.26953125, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.24804262816905975, "decoder_belief_target_nll": 7.497456073760986, "decoder_belief_top1_on_support": 0.25, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 3, "n": 256, "query_target_mass": 0.8925803303718567, "query_target_nll": 1.4899864196777344, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.7341902256011963, "probe_belief_target_nll": 2.1145966053009033, "probe_belief_top1_on_support": 0.83203125, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.6748705506324768, "decoder_belief_target_nll": 2.2791590690612793, "decoder_belief_top1_on_support": 0.765625, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 6, "n": 256, "query_target_mass": 0.8925803303718567, "query_target_nll": 1.4899864196777344, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.7341902256011963, "probe_belief_target_nll": 2.1145966053009033, "probe_belief_top1_on_support": 0.83203125, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.6748705506324768, "decoder_belief_target_nll": 2.2791590690612793, "decoder_belief_top1_on_support": 0.765625, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 9, "n": 256, "query_target_mass": 0.8925803303718567, "query_target_nll": 1.4899864196777344, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.7341902256011963, "probe_belief_target_nll": 2.1145966053009033, "probe_belief_top1_on_support": 0.83203125, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.6748705506324768, "decoder_belief_target_nll": 2.2791590690612793, "decoder_belief_top1_on_support": 0.765625, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A", "k": 12, "n": 256, "query_target_mass": 0.8925803303718567, "query_target_nll": 1.4899864196777344, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.18359375, "probe_belief_target_mass": 0.7341902256011963, "probe_belief_target_nll": 2.1145966053009033, "probe_belief_top1_on_support": 0.83203125, "mean_belief_support_size": 5.84375, "decoder_belief_target_mass": 0.6748705506324768, "decoder_belief_target_nll": 2.2791590690612793, "decoder_belief_top1_on_support": 0.765625, "mean_decoder_belief_support_size": 5.84375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 0, "n": 256, "query_target_mass": 0.4675194025039673, "query_target_nll": 2.4076099395751953, "query_top1_on_support": 0.41796875, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.053438495844602585, "probe_belief_target_nll": 14.691007614135742, "probe_belief_top1_on_support": 0.05859375, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.05178983137011528, "decoder_belief_target_nll": 14.607622146606445, "decoder_belief_top1_on_support": 0.03125, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 1, "n": 256, "query_target_mass": 0.5365063548088074, "query_target_nll": 3.6309354305267334, "query_top1_on_support": 0.546875, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.13562408089637756, "probe_belief_target_nll": 11.838923454284668, "probe_belief_top1_on_support": 0.1171875, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.13190221786499023, "decoder_belief_target_nll": 10.770563125610352, "decoder_belief_top1_on_support": 0.13671875, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 2, "n": 256, "query_target_mass": 0.6710456609725952, "query_target_nll": 3.090353488922119, "query_top1_on_support": 0.69921875, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.26412439346313477, "probe_belief_target_nll": 8.26401424407959, "probe_belief_top1_on_support": 0.2734375, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.2621672749519348, "decoder_belief_target_nll": 7.5432024002075195, "decoder_belief_top1_on_support": 0.2578125, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 3, "n": 256, "query_target_mass": 0.8925812840461731, "query_target_nll": 1.4727723598480225, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.7702730298042297, "probe_belief_target_nll": 2.0232126712799072, "probe_belief_top1_on_support": 0.83984375, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.7222208976745605, "decoder_belief_target_nll": 2.140378475189209, "decoder_belief_top1_on_support": 0.80078125, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 6, "n": 256, "query_target_mass": 0.8925812840461731, "query_target_nll": 1.4727723598480225, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.7702730298042297, "probe_belief_target_nll": 2.0232126712799072, "probe_belief_top1_on_support": 0.83984375, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.7222208976745605, "decoder_belief_target_nll": 2.140378475189209, "decoder_belief_top1_on_support": 0.80078125, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 9, "n": 256, "query_target_mass": 0.8925812840461731, "query_target_nll": 1.4727723598480225, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.7702730298042297, "probe_belief_target_nll": 2.0232126712799072, "probe_belief_top1_on_support": 0.83984375, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.7222208976745605, "decoder_belief_target_nll": 2.140378475189209, "decoder_belief_top1_on_support": 0.80078125, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "B", "k": 12, "n": 256, "query_target_mass": 0.8925812840461731, "query_target_nll": 1.4727723598480225, "query_top1_on_support": 0.9609375, "mean_query_support_size": 5.15234375, "probe_belief_target_mass": 0.7702730298042297, "probe_belief_target_nll": 2.0232126712799072, "probe_belief_top1_on_support": 0.83984375, "mean_belief_support_size": 5.77734375, "decoder_belief_target_mass": 0.7222208976745605, "decoder_belief_target_nll": 2.140378475189209, "decoder_belief_top1_on_support": 0.80078125, "mean_decoder_belief_support_size": 5.77734375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 0, "n": 256, "query_target_mass": 0.48800837993621826, "query_target_nll": 2.405651569366455, "query_top1_on_support": 0.484375, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.06515798717737198, "probe_belief_target_nll": 14.323688507080078, "probe_belief_top1_on_support": 0.078125, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.06708110868930817, "decoder_belief_target_nll": 14.262059211730957, "decoder_belief_top1_on_support": 0.078125, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 1, "n": 256, "query_target_mass": 0.4929470717906952, "query_target_nll": 3.3908491134643555, "query_top1_on_support": 0.49609375, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.1561374068260193, "probe_belief_target_nll": 12.225238800048828, "probe_belief_top1_on_support": 0.16796875, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.15605413913726807, "decoder_belief_target_nll": 11.164959907531738, "decoder_belief_top1_on_support": 0.18359375, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 2, "n": 256, "query_target_mass": 0.5318341851234436, "query_target_nll": 3.273054599761963, "query_top1_on_support": 0.53515625, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.24327345192432404, "probe_belief_target_nll": 9.204524040222168, "probe_belief_top1_on_support": 0.23828125, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.24055752158164978, "decoder_belief_target_nll": 8.251314163208008, "decoder_belief_top1_on_support": 0.2578125, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 3, "n": 256, "query_target_mass": 0.8041670322418213, "query_target_nll": 1.740978717803955, "query_top1_on_support": 0.91796875, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.8008775115013123, "probe_belief_target_nll": 1.868009328842163, "probe_belief_top1_on_support": 0.9140625, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.7430648803710938, "decoder_belief_target_nll": 2.018749713897705, "decoder_belief_top1_on_support": 0.8359375, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 6, "n": 256, "query_target_mass": 0.8041670322418213, "query_target_nll": 1.740978717803955, "query_top1_on_support": 0.91796875, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.8008775115013123, "probe_belief_target_nll": 1.868009328842163, "probe_belief_top1_on_support": 0.9140625, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.7430648803710938, "decoder_belief_target_nll": 2.018749713897705, "decoder_belief_top1_on_support": 0.8359375, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 9, "n": 256, "query_target_mass": 0.8041670322418213, "query_target_nll": 1.740978717803955, "query_top1_on_support": 0.91796875, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.8008775115013123, "probe_belief_target_nll": 1.868009328842163, "probe_belief_top1_on_support": 0.9140625, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.7430648803710938, "decoder_belief_target_nll": 2.018749713897705, "decoder_belief_top1_on_support": 0.8359375, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_PLUS_B", "k": 12, "n": 256, "query_target_mass": 0.8041670322418213, "query_target_nll": 1.740978717803955, "query_top1_on_support": 0.91796875, "mean_query_support_size": 5.3671875, "probe_belief_target_mass": 0.8008775115013123, "probe_belief_target_nll": 1.868009328842163, "probe_belief_top1_on_support": 0.9140625, "mean_belief_support_size": 5.484375, "decoder_belief_target_mass": 0.7430648803710938, "decoder_belief_target_nll": 2.018749713897705, "decoder_belief_top1_on_support": 0.8359375, "mean_decoder_belief_support_size": 5.484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 0, "n": 256, "query_target_mass": 0.47265106439590454, "query_target_nll": 12.126614570617676, "query_top1_on_support": 0.47265625, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.05878373980522156, "probe_belief_target_nll": 14.321184158325195, "probe_belief_top1_on_support": 0.05078125, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.05845846235752106, "decoder_belief_target_nll": 14.099020004272461, "decoder_belief_top1_on_support": 0.05859375, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 1, "n": 256, "query_target_mass": 0.5287535190582275, "query_target_nll": 7.039482593536377, "query_top1_on_support": 0.5234375, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.15872284770011902, "probe_belief_target_nll": 11.468069076538086, "probe_belief_top1_on_support": 0.171875, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.1548023223876953, "decoder_belief_target_nll": 10.47824478149414, "decoder_belief_top1_on_support": 0.140625, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 2, "n": 256, "query_target_mass": 0.5782934427261353, "query_target_nll": 4.97232723236084, "query_top1_on_support": 0.5859375, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.2868194878101349, "probe_belief_target_nll": 8.017508506774902, "probe_belief_top1_on_support": 0.29296875, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.28077012300491333, "decoder_belief_target_nll": 7.161227226257324, "decoder_belief_top1_on_support": 0.3125, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 3, "n": 256, "query_target_mass": 0.884506344795227, "query_target_nll": 1.2969026565551758, "query_top1_on_support": 0.9453125, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.7442786693572998, "probe_belief_target_nll": 2.016141891479492, "probe_belief_top1_on_support": 0.84765625, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.6885051727294922, "decoder_belief_target_nll": 2.187746047973633, "decoder_belief_top1_on_support": 0.78125, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 6, "n": 256, "query_target_mass": 0.884506344795227, "query_target_nll": 1.2969026565551758, "query_top1_on_support": 0.9453125, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.7442786693572998, "probe_belief_target_nll": 2.016141891479492, "probe_belief_top1_on_support": 0.84765625, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.6885051727294922, "decoder_belief_target_nll": 2.187746047973633, "decoder_belief_top1_on_support": 0.78125, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 9, "n": 256, "query_target_mass": 0.884506344795227, "query_target_nll": 1.2969026565551758, "query_top1_on_support": 0.9453125, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.7442786693572998, "probe_belief_target_nll": 2.016141891479492, "probe_belief_top1_on_support": 0.84765625, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.6885051727294922, "decoder_belief_target_nll": 2.187746047973633, "decoder_belief_top1_on_support": 0.78125, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 3, "query_type": "A_MINUS_B", "k": 12, "n": 256, "query_target_mass": 0.884506344795227, "query_target_nll": 1.2969026565551758, "query_top1_on_support": 0.9453125, "mean_query_support_size": 4.671875, "probe_belief_target_mass": 0.7442786693572998, "probe_belief_target_nll": 2.016141891479492, "probe_belief_top1_on_support": 0.84765625, "mean_belief_support_size": 5.6015625, "decoder_belief_target_mass": 0.6885051727294922, "decoder_belief_target_nll": 2.187746047973633, "decoder_belief_top1_on_support": 0.78125, "mean_decoder_belief_support_size": 5.6015625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 0, "n": 256, "query_target_mass": 0.2537221312522888, "query_target_nll": 2.4000742435455322, "query_top1_on_support": 0.26171875, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.029447171837091446, "probe_belief_target_nll": 14.838905334472656, "probe_belief_top1_on_support": 0.02734375, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.029841972514986992, "decoder_belief_target_nll": 14.726110458374023, "decoder_belief_top1_on_support": 0.0390625, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 1, "n": 256, "query_target_mass": 0.28137272596359253, "query_target_nll": 4.341009140014648, "query_top1_on_support": 0.28515625, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.07034552097320557, "probe_belief_target_nll": 13.303915023803711, "probe_belief_top1_on_support": 0.0703125, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.07203295081853867, "decoder_belief_target_nll": 12.075349807739258, "decoder_belief_top1_on_support": 0.078125, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 2, "n": 256, "query_target_mass": 0.3396562337875366, "query_target_nll": 4.190268516540527, "query_top1_on_support": 0.34765625, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.0884120762348175, "probe_belief_target_nll": 12.100830078125, "probe_belief_top1_on_support": 0.09375, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.08290617913007736, "decoder_belief_target_nll": 10.889278411865234, "decoder_belief_top1_on_support": 0.1015625, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 3, "n": 256, "query_target_mass": 0.3837561011314392, "query_target_nll": 4.076622009277344, "query_top1_on_support": 0.375, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.10573400557041168, "probe_belief_target_nll": 11.046472549438477, "probe_belief_top1_on_support": 0.1171875, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.10151016712188721, "decoder_belief_target_nll": 9.835972785949707, "decoder_belief_top1_on_support": 0.1171875, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 6, "n": 256, "query_target_mass": 0.6458101868629456, "query_target_nll": 1.3737127780914307, "query_top1_on_support": 0.7578125, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.4922741651535034, "probe_belief_target_nll": 2.2588436603546143, "probe_belief_top1_on_support": 0.609375, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.43104422092437744, "decoder_belief_target_nll": 2.3843908309936523, "decoder_belief_top1_on_support": 0.5703125, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 9, "n": 256, "query_target_mass": 0.6458101868629456, "query_target_nll": 1.3737127780914307, "query_top1_on_support": 0.7578125, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.4922741651535034, "probe_belief_target_nll": 2.2588436603546143, "probe_belief_top1_on_support": 0.609375, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.43104422092437744, "decoder_belief_target_nll": 2.3843908309936523, "decoder_belief_top1_on_support": 0.5703125, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A", "k": 12, "n": 256, "query_target_mass": 0.6458101868629456, "query_target_nll": 1.3737127780914307, "query_top1_on_support": 0.7578125, "mean_query_support_size": 2.78125, "probe_belief_target_mass": 0.4922741651535034, "probe_belief_target_nll": 2.2588436603546143, "probe_belief_top1_on_support": 0.609375, "mean_belief_support_size": 3.45703125, "decoder_belief_target_mass": 0.43104422092437744, "decoder_belief_target_nll": 2.3843908309936523, "decoder_belief_top1_on_support": 0.5703125, "mean_decoder_belief_support_size": 3.45703125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 0, "n": 256, "query_target_mass": 0.31064409017562866, "query_target_nll": 2.398022174835205, "query_top1_on_support": 0.30078125, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.034223511815071106, "probe_belief_target_nll": 15.159629821777344, "probe_belief_top1_on_support": 0.02734375, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.032324738800525665, "decoder_belief_target_nll": 15.171564102172852, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 1, "n": 256, "query_target_mass": 0.33975404500961304, "query_target_nll": 4.726959705352783, "query_top1_on_support": 0.3515625, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.04494212195277214, "probe_belief_target_nll": 14.63216781616211, "probe_belief_top1_on_support": 0.04296875, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.04523719102144241, "decoder_belief_target_nll": 13.099832534790039, "decoder_belief_top1_on_support": 0.046875, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 2, "n": 256, "query_target_mass": 0.3796362280845642, "query_target_nll": 4.4488630294799805, "query_top1_on_support": 0.37890625, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.07121298462152481, "probe_belief_target_nll": 12.76153564453125, "probe_belief_top1_on_support": 0.08203125, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.0715261697769165, "decoder_belief_target_nll": 11.315265655517578, "decoder_belief_top1_on_support": 0.078125, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 3, "n": 256, "query_target_mass": 0.4202496409416199, "query_target_nll": 4.6516618728637695, "query_top1_on_support": 0.453125, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.10124212503433228, "probe_belief_target_nll": 11.538277626037598, "probe_belief_top1_on_support": 0.1171875, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.09586749225854874, "decoder_belief_target_nll": 10.147336959838867, "decoder_belief_top1_on_support": 0.11328125, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 6, "n": 256, "query_target_mass": 0.6914786696434021, "query_target_nll": 1.5156967639923096, "query_top1_on_support": 0.7734375, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.47521302103996277, "probe_belief_target_nll": 2.4778449535369873, "probe_belief_top1_on_support": 0.59765625, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.42609384655952454, "decoder_belief_target_nll": 2.6048521995544434, "decoder_belief_top1_on_support": 0.55078125, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 9, "n": 256, "query_target_mass": 0.6914786696434021, "query_target_nll": 1.5156967639923096, "query_top1_on_support": 0.7734375, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.47521302103996277, "probe_belief_target_nll": 2.4778449535369873, "probe_belief_top1_on_support": 0.59765625, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.42609384655952454, "decoder_belief_target_nll": 2.6048521995544434, "decoder_belief_top1_on_support": 0.55078125, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "B", "k": 12, "n": 256, "query_target_mass": 0.6914786696434021, "query_target_nll": 1.5156967639923096, "query_top1_on_support": 0.7734375, "mean_query_support_size": 3.4140625, "probe_belief_target_mass": 0.47521302103996277, "probe_belief_target_nll": 2.4778449535369873, "probe_belief_top1_on_support": 0.59765625, "mean_belief_support_size": 4.03125, "decoder_belief_target_mass": 0.42609384655952454, "decoder_belief_target_nll": 2.6048521995544434, "decoder_belief_top1_on_support": 0.55078125, "mean_decoder_belief_support_size": 4.03125 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 0, "n": 256, "query_target_mass": 0.3297106921672821, "query_target_nll": 2.401933193206787, "query_top1_on_support": 0.3203125, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.033694278448820114, "probe_belief_target_nll": 15.114776611328125, "probe_belief_top1_on_support": 0.0390625, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.03472515195608139, "decoder_belief_target_nll": 14.892007827758789, "decoder_belief_top1_on_support": 0.04296875, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 1, "n": 256, "query_target_mass": 0.3369249105453491, "query_target_nll": 3.083573341369629, "query_top1_on_support": 0.33203125, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.051179319620132446, "probe_belief_target_nll": 13.797886848449707, "probe_belief_top1_on_support": 0.05078125, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.05086776614189148, "decoder_belief_target_nll": 12.417303085327148, "decoder_belief_top1_on_support": 0.06640625, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 2, "n": 256, "query_target_mass": 0.33210426568984985, "query_target_nll": 3.377225875854492, "query_top1_on_support": 0.3203125, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.06946124136447906, "probe_belief_target_nll": 12.156793594360352, "probe_belief_top1_on_support": 0.05859375, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.06832122802734375, "decoder_belief_target_nll": 10.767236709594727, "decoder_belief_top1_on_support": 0.078125, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 3, "n": 256, "query_target_mass": 0.32397377490997314, "query_target_nll": 3.409680128097534, "query_top1_on_support": 0.33984375, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.09265422821044922, "probe_belief_target_nll": 10.650568962097168, "probe_belief_top1_on_support": 0.09765625, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.0919775441288948, "decoder_belief_target_nll": 9.455127716064453, "decoder_belief_top1_on_support": 0.1015625, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 6, "n": 256, "query_target_mass": 0.5550178289413452, "query_target_nll": 1.8201589584350586, "query_top1_on_support": 0.66015625, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.44766730070114136, "probe_belief_target_nll": 2.51962947845459, "probe_belief_top1_on_support": 0.57421875, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.3893982470035553, "decoder_belief_target_nll": 2.625704526901245, "decoder_belief_top1_on_support": 0.50390625, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 9, "n": 256, "query_target_mass": 0.5550178289413452, "query_target_nll": 1.8201589584350586, "query_top1_on_support": 0.66015625, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.44766730070114136, "probe_belief_target_nll": 2.51962947845459, "probe_belief_top1_on_support": 0.57421875, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.3893982470035553, "decoder_belief_target_nll": 2.625704526901245, "decoder_belief_top1_on_support": 0.50390625, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_PLUS_B", "k": 12, "n": 256, "query_target_mass": 0.5550178289413452, "query_target_nll": 1.8201589584350586, "query_top1_on_support": 0.66015625, "mean_query_support_size": 3.6171875, "probe_belief_target_mass": 0.44766730070114136, "probe_belief_target_nll": 2.51962947845459, "probe_belief_top1_on_support": 0.57421875, "mean_belief_support_size": 3.97265625, "decoder_belief_target_mass": 0.3893982470035553, "decoder_belief_target_nll": 2.625704526901245, "decoder_belief_top1_on_support": 0.50390625, "mean_decoder_belief_support_size": 3.97265625 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 0, "n": 256, "query_target_mass": 0.3124987781047821, "query_target_nll": 12.642045021057129, "query_top1_on_support": 0.3125, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.03171379119157791, "probe_belief_target_nll": 14.804145812988281, "probe_belief_top1_on_support": 0.0390625, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.03221919387578964, "decoder_belief_target_nll": 14.665407180786133, "decoder_belief_top1_on_support": 0.03125, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 1, "n": 256, "query_target_mass": 0.3024776577949524, "query_target_nll": 7.99769401550293, "query_top1_on_support": 0.31640625, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.04385068640112877, "probe_belief_target_nll": 13.85090446472168, "probe_belief_top1_on_support": 0.0546875, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.04532470554113388, "decoder_belief_target_nll": 12.4837646484375, "decoder_belief_top1_on_support": 0.0546875, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 2, "n": 256, "query_target_mass": 0.33139804005622864, "query_target_nll": 6.402298927307129, "query_top1_on_support": 0.32421875, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.06954978406429291, "probe_belief_target_nll": 12.337353706359863, "probe_belief_top1_on_support": 0.06640625, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.0717373862862587, "decoder_belief_target_nll": 11.030841827392578, "decoder_belief_top1_on_support": 0.0859375, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 3, "n": 256, "query_target_mass": 0.3156290054321289, "query_target_nll": 5.583003997802734, "query_top1_on_support": 0.296875, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.08076262474060059, "probe_belief_target_nll": 11.191756248474121, "probe_belief_top1_on_support": 0.09375, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.07859646528959274, "decoder_belief_target_nll": 9.905939102172852, "decoder_belief_top1_on_support": 0.1015625, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 6, "n": 256, "query_target_mass": 0.6181237697601318, "query_target_nll": 1.584895372390747, "query_top1_on_support": 0.6953125, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.461159884929657, "probe_belief_target_nll": 2.4052951335906982, "probe_belief_top1_on_support": 0.58984375, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.40318524837493896, "decoder_belief_target_nll": 2.5739731788635254, "decoder_belief_top1_on_support": 0.52734375, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 9, "n": 256, "query_target_mass": 0.6181237697601318, "query_target_nll": 1.584895372390747, "query_top1_on_support": 0.6953125, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.461159884929657, "probe_belief_target_nll": 2.4052951335906982, "probe_belief_top1_on_support": 0.58984375, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.40318524837493896, "decoder_belief_target_nll": 2.5739731788635254, "decoder_belief_top1_on_support": 0.52734375, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 6, "query_type": "A_MINUS_B", "k": 12, "n": 256, "query_target_mass": 0.6181237697601318, "query_target_nll": 1.584895372390747, "query_top1_on_support": 0.6953125, "mean_query_support_size": 3.2265625, "probe_belief_target_mass": 0.461159884929657, "probe_belief_target_nll": 2.4052951335906982, "probe_belief_top1_on_support": 0.58984375, "mean_belief_support_size": 3.76171875, "decoder_belief_target_mass": 0.40318524837493896, "decoder_belief_target_nll": 2.5739731788635254, "decoder_belief_top1_on_support": 0.52734375, "mean_decoder_belief_support_size": 3.76171875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 0, "n": 256, "query_target_mass": 0.2189180701971054, "query_target_nll": 2.403144359588623, "query_top1_on_support": 0.203125, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.02176831103861332, "probe_belief_target_nll": 15.062162399291992, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.02052629366517067, "decoder_belief_target_nll": 14.952841758728027, "decoder_belief_top1_on_support": 0.01953125, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 1, "n": 256, "query_target_mass": 0.24017047882080078, "query_target_nll": 4.404057502746582, "query_top1_on_support": 0.2265625, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.03532690554857254, "probe_belief_target_nll": 13.88127326965332, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.03599216043949127, "decoder_belief_target_nll": 12.639251708984375, "decoder_belief_top1_on_support": 0.0390625, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 2, "n": 256, "query_target_mass": 0.25950509309768677, "query_target_nll": 4.989519119262695, "query_top1_on_support": 0.234375, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.033728502690792084, "probe_belief_target_nll": 13.393903732299805, "probe_belief_top1_on_support": 0.0390625, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.033231787383556366, "decoder_belief_target_nll": 11.918424606323242, "decoder_belief_top1_on_support": 0.02734375, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 3, "n": 256, "query_target_mass": 0.26300016045570374, "query_target_nll": 5.260317802429199, "query_top1_on_support": 0.234375, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.042436450719833374, "probe_belief_target_nll": 12.708333015441895, "probe_belief_top1_on_support": 0.046875, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.04270734637975693, "decoder_belief_target_nll": 11.199536323547363, "decoder_belief_top1_on_support": 0.04296875, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 6, "n": 256, "query_target_mass": 0.3037038743495941, "query_target_nll": 4.681426048278809, "query_top1_on_support": 0.29296875, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.06512287259101868, "probe_belief_target_nll": 10.026708602905273, "probe_belief_top1_on_support": 0.0703125, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.0631619542837143, "decoder_belief_target_nll": 8.822229385375977, "decoder_belief_top1_on_support": 0.0703125, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 9, "n": 256, "query_target_mass": 0.5323264598846436, "query_target_nll": 1.5239917039871216, "query_top1_on_support": 0.609375, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.29518723487854004, "probe_belief_target_nll": 2.7319247722625732, "probe_belief_top1_on_support": 0.37109375, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.2538345456123352, "decoder_belief_target_nll": 2.854013442993164, "decoder_belief_top1_on_support": 0.34375, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A", "k": 12, "n": 256, "query_target_mass": 0.5323264598846436, "query_target_nll": 1.5239917039871216, "query_top1_on_support": 0.609375, "mean_query_support_size": 2.40625, "probe_belief_target_mass": 0.29518723487854004, "probe_belief_target_nll": 2.7319247722625732, "probe_belief_top1_on_support": 0.37109375, "mean_belief_support_size": 2.73828125, "decoder_belief_target_mass": 0.2538345456123352, "decoder_belief_target_nll": 2.854013442993164, "decoder_belief_top1_on_support": 0.34375, "mean_decoder_belief_support_size": 2.73828125 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 0, "n": 256, "query_target_mass": 0.19176506996154785, "query_target_nll": 2.408548593521118, "query_top1_on_support": 0.1796875, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.02568979188799858, "probe_belief_target_nll": 14.835430145263672, "probe_belief_top1_on_support": 0.0390625, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.02445192262530327, "decoder_belief_target_nll": 14.834799766540527, "decoder_belief_top1_on_support": 0.015625, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 1, "n": 256, "query_target_mass": 0.19427445530891418, "query_target_nll": 4.674410820007324, "query_top1_on_support": 0.18359375, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.025338197126984596, "probe_belief_target_nll": 14.23746395111084, "probe_belief_top1_on_support": 0.02734375, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.025096222758293152, "decoder_belief_target_nll": 12.807504653930664, "decoder_belief_top1_on_support": 0.03515625, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 2, "n": 256, "query_target_mass": 0.20838649570941925, "query_target_nll": 4.709427833557129, "query_top1_on_support": 0.21875, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.029847029596567154, "probe_belief_target_nll": 12.95082950592041, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.03009396232664585, "decoder_belief_target_nll": 11.502788543701172, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 3, "n": 256, "query_target_mass": 0.22199100255966187, "query_target_nll": 4.91827917098999, "query_top1_on_support": 0.22265625, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.033362485468387604, "probe_belief_target_nll": 12.343295097351074, "probe_belief_top1_on_support": 0.02734375, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.037119925022125244, "decoder_belief_target_nll": 10.859235763549805, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 6, "n": 256, "query_target_mass": 0.2706854045391083, "query_target_nll": 4.909719467163086, "query_top1_on_support": 0.30078125, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.04250621423125267, "probe_belief_target_nll": 10.613279342651367, "probe_belief_top1_on_support": 0.046875, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.03965207189321518, "decoder_belief_target_nll": 9.37588119506836, "decoder_belief_top1_on_support": 0.03515625, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 9, "n": 256, "query_target_mass": 0.5129725337028503, "query_target_nll": 1.5279830694198608, "query_top1_on_support": 0.60546875, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.2984168231487274, "probe_belief_target_nll": 2.6530539989471436, "probe_belief_top1_on_support": 0.37109375, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.2484404593706131, "decoder_belief_target_nll": 2.770081043243408, "decoder_belief_top1_on_support": 0.32421875, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "B", "k": 12, "n": 256, "query_target_mass": 0.5129725337028503, "query_target_nll": 1.5279830694198608, "query_top1_on_support": 0.60546875, "mean_query_support_size": 2.1171875, "probe_belief_target_mass": 0.2984168231487274, "probe_belief_target_nll": 2.6530539989471436, "probe_belief_top1_on_support": 0.37109375, "mean_belief_support_size": 2.37109375, "decoder_belief_target_mass": 0.2484404593706131, "decoder_belief_target_nll": 2.770081043243408, "decoder_belief_top1_on_support": 0.32421875, "mean_decoder_belief_support_size": 2.37109375 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 0, "n": 256, "query_target_mass": 0.22884848713874817, "query_target_nll": 2.4010908603668213, "query_top1_on_support": 0.22265625, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.022631926462054253, "probe_belief_target_nll": 15.036796569824219, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.022296041250228882, "decoder_belief_target_nll": 14.99567985534668, "decoder_belief_top1_on_support": 0.01953125, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 1, "n": 256, "query_target_mass": 0.23531559109687805, "query_target_nll": 3.1042914390563965, "query_top1_on_support": 0.23828125, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.030372777953743935, "probe_belief_target_nll": 14.18662166595459, "probe_belief_top1_on_support": 0.03515625, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.030792664736509323, "decoder_belief_target_nll": 12.79656982421875, "decoder_belief_top1_on_support": 0.03515625, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 2, "n": 256, "query_target_mass": 0.21438074111938477, "query_target_nll": 3.502664566040039, "query_top1_on_support": 0.22265625, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.026242781430482864, "probe_belief_target_nll": 13.162332534790039, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.026975730434060097, "decoder_belief_target_nll": 11.707018852233887, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 3, "n": 256, "query_target_mass": 0.22726237773895264, "query_target_nll": 3.489077091217041, "query_top1_on_support": 0.24609375, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.0404343456029892, "probe_belief_target_nll": 11.829819679260254, "probe_belief_top1_on_support": 0.0546875, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.04169577732682228, "decoder_belief_target_nll": 10.439943313598633, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 6, "n": 256, "query_target_mass": 0.2277146875858307, "query_target_nll": 3.3670482635498047, "query_top1_on_support": 0.2421875, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.05805189907550812, "probe_belief_target_nll": 9.840456008911133, "probe_belief_top1_on_support": 0.06640625, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.05575031042098999, "decoder_belief_target_nll": 8.586715698242188, "decoder_belief_top1_on_support": 0.07421875, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 9, "n": 256, "query_target_mass": 0.4036925733089447, "query_target_nll": 1.9036612510681152, "query_top1_on_support": 0.484375, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.29163575172424316, "probe_belief_target_nll": 2.7613840103149414, "probe_belief_top1_on_support": 0.33984375, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.24818193912506104, "decoder_belief_target_nll": 2.8561925888061523, "decoder_belief_top1_on_support": 0.27734375, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_PLUS_B", "k": 12, "n": 256, "query_target_mass": 0.4036925733089447, "query_target_nll": 1.9036612510681152, "query_top1_on_support": 0.484375, "mean_query_support_size": 2.50390625, "probe_belief_target_mass": 0.29163575172424316, "probe_belief_target_nll": 2.7613840103149414, "probe_belief_top1_on_support": 0.33984375, "mean_belief_support_size": 2.71875, "decoder_belief_target_mass": 0.24818193912506104, "decoder_belief_target_nll": 2.8561925888061523, "decoder_belief_top1_on_support": 0.27734375, "mean_decoder_belief_support_size": 2.71875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 0, "n": 256, "query_target_mass": 0.18359753489494324, "query_target_nll": 13.08549690246582, "query_top1_on_support": 0.18359375, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.01667170785367489, "probe_belief_target_nll": 15.410394668579102, "probe_belief_top1_on_support": 0.015625, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.01728162355720997, "decoder_belief_target_nll": 15.37843132019043, "decoder_belief_top1_on_support": 0.015625, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 1, "n": 256, "query_target_mass": 0.201188862323761, "query_target_nll": 7.444463729858398, "query_top1_on_support": 0.21875, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.02151026576757431, "probe_belief_target_nll": 13.919490814208984, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.021015550941228867, "decoder_belief_target_nll": 12.612281799316406, "decoder_belief_top1_on_support": 0.01171875, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 2, "n": 256, "query_target_mass": 0.23479968309402466, "query_target_nll": 6.397564888000488, "query_top1_on_support": 0.25390625, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.023283354938030243, "probe_belief_target_nll": 12.92481803894043, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.024315908551216125, "decoder_belief_target_nll": 11.461416244506836, "decoder_belief_top1_on_support": 0.02734375, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 3, "n": 256, "query_target_mass": 0.23523691296577454, "query_target_nll": 5.612215042114258, "query_top1_on_support": 0.21875, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.02535780891776085, "probe_belief_target_nll": 12.569501876831055, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.027526147663593292, "decoder_belief_target_nll": 11.028861999511719, "decoder_belief_top1_on_support": 0.03125, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 6, "n": 256, "query_target_mass": 0.22928373515605927, "query_target_nll": 4.747409820556641, "query_top1_on_support": 0.23046875, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.0585569329559803, "probe_belief_target_nll": 10.283754348754883, "probe_belief_top1_on_support": 0.0625, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.05095735564827919, "decoder_belief_target_nll": 8.948102951049805, "decoder_belief_top1_on_support": 0.0625, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 9, "n": 256, "query_target_mass": 0.46213194727897644, "query_target_nll": 1.7293157577514648, "query_top1_on_support": 0.58203125, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.29340070486068726, "probe_belief_target_nll": 2.662372350692749, "probe_belief_top1_on_support": 0.375, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.25131410360336304, "decoder_belief_target_nll": 2.736961841583252, "decoder_belief_top1_on_support": 0.31640625, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 9, "query_type": "A_MINUS_B", "k": 12, "n": 256, "query_target_mass": 0.46213194727897644, "query_target_nll": 1.7293157577514648, "query_top1_on_support": 0.58203125, "mean_query_support_size": 2.35546875, "probe_belief_target_mass": 0.29340070486068726, "probe_belief_target_nll": 2.662372350692749, "probe_belief_top1_on_support": 0.375, "mean_belief_support_size": 2.48046875, "decoder_belief_target_mass": 0.25131410360336304, "decoder_belief_target_nll": 2.736961841583252, "decoder_belief_top1_on_support": 0.31640625, "mean_decoder_belief_support_size": 2.48046875 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 0, "n": 256, "query_target_mass": 0.1736900806427002, "query_target_nll": 2.397264003753662, "query_top1_on_support": 0.1328125, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.017749248072504997, "probe_belief_target_nll": 14.949369430541992, "probe_belief_top1_on_support": 0.015625, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.017574883997440338, "decoder_belief_target_nll": 14.907354354858398, "decoder_belief_top1_on_support": 0.03125, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 1, "n": 256, "query_target_mass": 0.18890106678009033, "query_target_nll": 4.386326313018799, "query_top1_on_support": 0.19140625, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.028723519295454025, "probe_belief_target_nll": 13.911731719970703, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.029775263741612434, "decoder_belief_target_nll": 12.666749000549316, "decoder_belief_top1_on_support": 0.0390625, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 2, "n": 256, "query_target_mass": 0.18905451893806458, "query_target_nll": 5.07179594039917, "query_top1_on_support": 0.21875, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.019826740026474, "probe_belief_target_nll": 13.524459838867188, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.01968027837574482, "decoder_belief_target_nll": 12.021086692810059, "decoder_belief_top1_on_support": 0.01953125, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 3, "n": 256, "query_target_mass": 0.1813749074935913, "query_target_nll": 5.428088188171387, "query_top1_on_support": 0.18359375, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.024649744853377342, "probe_belief_target_nll": 12.564220428466797, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.024539537727832794, "decoder_belief_target_nll": 11.033712387084961, "decoder_belief_top1_on_support": 0.02734375, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 6, "n": 256, "query_target_mass": 0.19359469413757324, "query_target_nll": 5.1394853591918945, "query_top1_on_support": 0.21875, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.022782038897275925, "probe_belief_target_nll": 11.376535415649414, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.02131274901330471, "decoder_belief_target_nll": 9.88347053527832, "decoder_belief_top1_on_support": 0.01953125, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 9, "n": 256, "query_target_mass": 0.25948864221572876, "query_target_nll": 4.1801018714904785, "query_top1_on_support": 0.3046875, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.05178367346525192, "probe_belief_target_nll": 9.429905891418457, "probe_belief_top1_on_support": 0.05859375, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.049113739281892776, "decoder_belief_target_nll": 8.359269142150879, "decoder_belief_top1_on_support": 0.05859375, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A", "k": 12, "n": 256, "query_target_mass": 0.43144625425338745, "query_target_nll": 1.5848922729492188, "query_top1_on_support": 0.48046875, "mean_query_support_size": 1.90625, "probe_belief_target_mass": 0.2171134650707245, "probe_belief_target_nll": 2.8624887466430664, "probe_belief_top1_on_support": 0.26953125, "mean_belief_support_size": 2.21484375, "decoder_belief_target_mass": 0.176906019449234, "decoder_belief_target_nll": 3.007249116897583, "decoder_belief_top1_on_support": 0.2109375, "mean_decoder_belief_support_size": 2.21484375 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 0, "n": 256, "query_target_mass": 0.1527685523033142, "query_target_nll": 2.4025843143463135, "query_top1_on_support": 0.140625, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.013638444244861603, "probe_belief_target_nll": 15.163543701171875, "probe_belief_top1_on_support": 0.02734375, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.013513337820768356, "decoder_belief_target_nll": 15.147950172424316, "decoder_belief_top1_on_support": 0.01171875, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 1, "n": 256, "query_target_mass": 0.15545663237571716, "query_target_nll": 4.171543121337891, "query_top1_on_support": 0.15625, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.01518799178302288, "probe_belief_target_nll": 14.103096961975098, "probe_belief_top1_on_support": 0.01953125, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.013587623834609985, "decoder_belief_target_nll": 12.85125732421875, "decoder_belief_top1_on_support": 0.0078125, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 2, "n": 256, "query_target_mass": 0.17336106300354004, "query_target_nll": 4.815938949584961, "query_top1_on_support": 0.17578125, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.01660696417093277, "probe_belief_target_nll": 12.935226440429688, "probe_belief_top1_on_support": 0.01171875, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.01743243634700775, "decoder_belief_target_nll": 11.477195739746094, "decoder_belief_top1_on_support": 0.01171875, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 3, "n": 256, "query_target_mass": 0.1729828119277954, "query_target_nll": 5.050936698913574, "query_top1_on_support": 0.18359375, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.013660077005624771, "probe_belief_target_nll": 12.497636795043945, "probe_belief_top1_on_support": 0.0078125, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.012630825862288475, "decoder_belief_target_nll": 10.968015670776367, "decoder_belief_top1_on_support": 0.0078125, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 6, "n": 256, "query_target_mass": 0.15856410562992096, "query_target_nll": 5.588208198547363, "query_top1_on_support": 0.16015625, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.019017238169908524, "probe_belief_target_nll": 11.665430068969727, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.019833872094750404, "decoder_belief_target_nll": 10.324474334716797, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 9, "n": 256, "query_target_mass": 0.23676224052906036, "query_target_nll": 4.098996639251709, "query_top1_on_support": 0.24609375, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.050763726234436035, "probe_belief_target_nll": 9.281917572021484, "probe_belief_top1_on_support": 0.04296875, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.04463934898376465, "decoder_belief_target_nll": 8.206031799316406, "decoder_belief_top1_on_support": 0.04296875, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "B", "k": 12, "n": 256, "query_target_mass": 0.427474707365036, "query_target_nll": 1.586342692375183, "query_top1_on_support": 0.49609375, "mean_query_support_size": 1.6796875, "probe_belief_target_mass": 0.20335501432418823, "probe_belief_target_nll": 2.820639133453369, "probe_belief_top1_on_support": 0.29296875, "mean_belief_support_size": 1.84765625, "decoder_belief_target_mass": 0.17177902162075043, "decoder_belief_target_nll": 2.9344096183776855, "decoder_belief_top1_on_support": 0.21875, "mean_decoder_belief_support_size": 1.84765625 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 0, "n": 256, "query_target_mass": 0.17636975646018982, "query_target_nll": 2.402407169342041, "query_top1_on_support": 0.15625, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.025971492752432823, "probe_belief_target_nll": 14.767010688781738, "probe_belief_top1_on_support": 0.046875, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.025011960417032242, "decoder_belief_target_nll": 14.618162155151367, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 1, "n": 256, "query_target_mass": 0.18688786029815674, "query_target_nll": 3.165393829345703, "query_top1_on_support": 0.1640625, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.016020525246858597, "probe_belief_target_nll": 14.306239128112793, "probe_belief_top1_on_support": 0.02734375, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.014762045815587044, "decoder_belief_target_nll": 12.935591697692871, "decoder_belief_top1_on_support": 0.0078125, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 2, "n": 256, "query_target_mass": 0.17643144726753235, "query_target_nll": 3.4669384956359863, "query_top1_on_support": 0.1640625, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.020933285355567932, "probe_belief_target_nll": 13.467761039733887, "probe_belief_top1_on_support": 0.0234375, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.0205984003841877, "decoder_belief_target_nll": 12.012868881225586, "decoder_belief_top1_on_support": 0.015625, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 3, "n": 256, "query_target_mass": 0.17394909262657166, "query_target_nll": 3.553793430328369, "query_top1_on_support": 0.16015625, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.015167457982897758, "probe_belief_target_nll": 13.249154090881348, "probe_belief_top1_on_support": 0.0078125, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.013838893733918667, "decoder_belief_target_nll": 11.61631965637207, "decoder_belief_top1_on_support": 0.0078125, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 6, "n": 256, "query_target_mass": 0.17384961247444153, "query_target_nll": 3.429617404937744, "query_top1_on_support": 0.171875, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.025776579976081848, "probe_belief_target_nll": 11.520524978637695, "probe_belief_top1_on_support": 0.03125, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.025523023679852486, "decoder_belief_target_nll": 10.048135757446289, "decoder_belief_top1_on_support": 0.03125, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 9, "n": 256, "query_target_mass": 0.17819549143314362, "query_target_nll": 3.365389347076416, "query_top1_on_support": 0.18359375, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.04121079668402672, "probe_belief_target_nll": 9.490903854370117, "probe_belief_top1_on_support": 0.05859375, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.0393759049475193, "decoder_belief_target_nll": 8.310565948486328, "decoder_belief_top1_on_support": 0.05859375, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_PLUS_B", "k": 12, "n": 256, "query_target_mass": 0.316808819770813, "query_target_nll": 1.9047154188156128, "query_top1_on_support": 0.421875, "mean_query_support_size": 1.93359375, "probe_belief_target_mass": 0.20264989137649536, "probe_belief_target_nll": 2.89009165763855, "probe_belief_top1_on_support": 0.28515625, "mean_belief_support_size": 2.0, "decoder_belief_target_mass": 0.17874881625175476, "decoder_belief_target_nll": 2.953000068664551, "decoder_belief_top1_on_support": 0.26171875, "mean_decoder_belief_support_size": 2.0 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 0, "n": 256, "query_target_mass": 0.21483832597732544, "query_target_nll": 12.640419006347656, "query_top1_on_support": 0.21484375, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.01995840109884739, "probe_belief_target_nll": 14.751235961914062, "probe_belief_top1_on_support": 0.015625, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.020688673481345177, "decoder_belief_target_nll": 14.699422836303711, "decoder_belief_top1_on_support": 0.0234375, "mean_decoder_belief_support_size": 2.17578125 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 1, "n": 256, "query_target_mass": 0.18193310499191284, "query_target_nll": 7.649951934814453, "query_top1_on_support": 0.171875, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.016352780163288116, "probe_belief_target_nll": 14.765544891357422, "probe_belief_top1_on_support": 0.01171875, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.017365511506795883, "decoder_belief_target_nll": 13.288411140441895, "decoder_belief_top1_on_support": 0.015625, "mean_decoder_belief_support_size": 2.17578125 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 2, "n": 256, "query_target_mass": 0.16097809374332428, "query_target_nll": 6.30814266204834, "query_top1_on_support": 0.16796875, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.015201512724161148, "probe_belief_target_nll": 13.740371704101562, "probe_belief_top1_on_support": 0.0, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.01753023825585842, "decoder_belief_target_nll": 12.121868133544922, "decoder_belief_top1_on_support": 0.015625, "mean_decoder_belief_support_size": 2.17578125 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 3, "n": 256, "query_target_mass": 0.16724663972854614, "query_target_nll": 5.528878211975098, "query_top1_on_support": 0.171875, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.013871900737285614, "probe_belief_target_nll": 12.835823059082031, "probe_belief_top1_on_support": 0.015625, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.01455758512020111, "decoder_belief_target_nll": 11.266945838928223, "decoder_belief_top1_on_support": 0.00390625, "mean_decoder_belief_support_size": 2.17578125 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 6, "n": 256, "query_target_mass": 0.1812504529953003, "query_target_nll": 4.46818208694458, "query_top1_on_support": 0.19921875, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.019611423835158348, "probe_belief_target_nll": 11.614599227905273, "probe_belief_top1_on_support": 0.03515625, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.018406927585601807, "decoder_belief_target_nll": 10.14895248413086, "decoder_belief_top1_on_support": 0.01953125, "mean_decoder_belief_support_size": 2.17578125 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 9, "n": 256, "query_target_mass": 0.18297553062438965, "query_target_nll": 4.1812567710876465, "query_top1_on_support": 0.17578125, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.04953975975513458, "probe_belief_target_nll": 9.25680923461914, "probe_belief_top1_on_support": 0.05859375, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.0408143475651741, "decoder_belief_target_nll": 8.086263656616211, "decoder_belief_top1_on_support": 0.0546875, "mean_decoder_belief_support_size": 2.17578125 }, { "model": "dense", "supervision": "sparse_belief", "length": 12, "query_type": "A_MINUS_B", "k": 12, "n": 256, "query_target_mass": 0.37632375955581665, "query_target_nll": 1.8401546478271484, "query_top1_on_support": 0.48828125, "mean_query_support_size": 1.95703125, "probe_belief_target_mass": 0.19924984872341156, "probe_belief_target_nll": 2.9911861419677734, "probe_belief_top1_on_support": 0.30078125, "mean_belief_support_size": 2.17578125, "decoder_belief_target_mass": 0.1710464358329773, "decoder_belief_target_nll": 3.0492897033691406, "decoder_belief_top1_on_support": 0.25, "mean_decoder_belief_support_size": 2.17578125 } ] } ], "checkpoints": [ "large_artifacts/dense_supervision_ladder/checkpoints/pilot_sparse_belief_mod11/checkpoint_final.pt" ] }