{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 11918,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "affinity_loss": 0.0214095051237382,
      "emkd_kd_loss": 2.554436580543406,
      "epoch": 0.0041956648293151105,
      "grad_norm": 3.387537717819214,
      "hard_loss": 1.674174085700506,
      "joint_kd_loss": 186.8290387225151,
      "learning_rate": 1.3687150837988828e-06,
      "loss": 107.680771484375,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 0.8516592668612725,
      "sre_geom_loss": 0.0071015077506353915,
      "sre_kd_loss": 371.10364186763763,
      "sre_logit_loss": 369.79138473510744,
      "sre_span_loss": 0.9571819393336773,
      "step": 50,
      "vision_kd_loss": 2.1286069425940513,
      "vision_logit_loss": 6.373477241396904
    },
    {
      "affinity_loss": 0.021203489053295926,
      "emkd_kd_loss": 2.587736677415669,
      "epoch": 0.008391329658630221,
      "grad_norm": 3.714298963546753,
      "hard_loss": 1.4310583000965562,
      "joint_kd_loss": 193.05748296976088,
      "learning_rate": 2.7653631284916204e-06,
      "loss": 101.3470703125,
      "matched_vision_samples": 0.95875,
      "response_kd_loss": 0.9188291165736859,
      "sre_geom_loss": 0.006304558528127018,
      "sre_kd_loss": 383.52722895145416,
      "sre_logit_loss": 382.3566454768181,
      "sre_span_loss": 0.8553563417494296,
      "step": 100,
      "vision_kd_loss": 2.128322115242481,
      "vision_logit_loss": 6.3929395627975465
    },
    {
      "affinity_loss": 0.021174087732797487,
      "emkd_kd_loss": 3.400079771503806,
      "epoch": 0.01258699448794533,
      "grad_norm": 18.12430191040039,
      "hard_loss": 1.5758969279733719,
      "joint_kd_loss": 138.35248623013496,
      "learning_rate": 4.1620111731843575e-06,
      "loss": 75.1925439453125,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 2.1953509214154714,
      "sre_geom_loss": 0.005298230306261757,
      "sre_kd_loss": 273.30489230751994,
      "sre_logit_loss": 272.4212029838562,
      "sre_span_loss": 0.6187775195389986,
      "step": 150,
      "vision_kd_loss": 2.3024043123424054,
      "vision_logit_loss": 7.092208485007286
    },
    {
      "affinity_loss": 0.05857324752956629,
      "emkd_kd_loss": 6.21337258502841,
      "epoch": 0.016782659317260442,
      "grad_norm": 2.1675057411193848,
      "hard_loss": 2.479635674666788,
      "joint_kd_loss": 19.732819819897415,
      "learning_rate": 5.558659217877096e-06,
      "loss": 11.504749755859375,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 4.807634055643575,
      "sre_geom_loss": 0.004481295977758322,
      "sre_kd_loss": 33.25226692542434,
      "sre_logit_loss": 32.59640333645046,
      "sre_span_loss": 0.4317988568544388,
      "step": 200,
      "vision_kd_loss": 3.809555558860302,
      "vision_logit_loss": 9.380897489786149
    },
    {
      "affinity_loss": 0.02941681042779237,
      "emkd_kd_loss": 2.9363309904187918,
      "epoch": 0.02097832414657555,
      "grad_norm": 3.2717490196228027,
      "hard_loss": 1.3216460885066772,
      "joint_kd_loss": 9.514868202656508,
      "learning_rate": 6.9553072625698325e-06,
      "loss": 4.879202270507813,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.6233197513426056,
      "sre_geom_loss": 0.004115921913090688,
      "sre_kd_loss": 16.093405328802763,
      "sre_logit_loss": 15.653532103174365,
      "sre_span_loss": 0.23407704018056394,
      "step": 250,
      "vision_kd_loss": 2.1246711164712906,
      "vision_logit_loss": 5.557003408670425
    },
    {
      "affinity_loss": 0.018218466729158535,
      "emkd_kd_loss": 2.371727115511894,
      "epoch": 0.02517398897589066,
      "grad_norm": 0.7552331686019897,
      "hard_loss": 1.210472868436773,
      "joint_kd_loss": 7.37300005659461,
      "learning_rate": 8.35195530726257e-06,
      "loss": 4.292313232421875,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.339163877714459,
      "sre_geom_loss": 0.0035324360415923193,
      "sre_kd_loss": 12.374273081030697,
      "sre_logit_loss": 12.013132381555625,
      "sre_span_loss": 0.18451884239912034,
      "step": 300,
      "vision_kd_loss": 1.7021451790630817,
      "vision_logit_loss": 4.986734040379524
    },
    {
      "affinity_loss": 0.015175756161333993,
      "emkd_kd_loss": 2.154384836405516,
      "epoch": 0.029369653805205773,
      "grad_norm": 1.235630750656128,
      "hard_loss": 1.0816440842110024,
      "joint_kd_loss": 6.238865802139044,
      "learning_rate": 9.748603351955308e-06,
      "loss": 3.68009765625,
      "matched_vision_samples": 0.95875,
      "response_kd_loss": 1.2668564593303744,
      "sre_geom_loss": 0.0035200351330126976,
      "sre_kd_loss": 10.32334672609344,
      "sre_logit_loss": 9.97958446983248,
      "sre_span_loss": 0.1677605168800801,
      "step": 350,
      "vision_kd_loss": 1.5209566101431846,
      "vision_logit_loss": 4.56625082463026
    },
    {
      "affinity_loss": 0.013026055314112455,
      "emkd_kd_loss": 1.983774196729064,
      "epoch": 0.033565318634520884,
      "grad_norm": 0.8168343305587769,
      "hard_loss": 1.0906158754891657,
      "joint_kd_loss": 5.953382775261998,
      "learning_rate": 9.999689624590088e-06,
      "loss": 3.5844671630859377,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.214701554281437,
      "sre_geom_loss": 0.0032824266041092186,
      "sre_kd_loss": 9.922991422209888,
      "sre_logit_loss": 9.604897236800753,
      "sre_span_loss": 0.15397282694466413,
      "step": 400,
      "vision_kd_loss": 1.3764234215021134,
      "vision_logit_loss": 4.203088153004646
    },
    {
      "affinity_loss": 0.011289949677884579,
      "emkd_kd_loss": 1.8461968129873276,
      "epoch": 0.03776098346383599,
      "grad_norm": 0.5271986126899719,
      "hard_loss": 1.0286246411150932,
      "joint_kd_loss": 6.64206777766347,
      "learning_rate": 9.998471080090999e-06,
      "loss": 3.352886962890625,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.1470870421752262,
      "sre_geom_loss": 0.003361915005677929,
      "sre_kd_loss": 11.437938760425896,
      "sre_logit_loss": 11.13104612413561,
      "sre_span_loss": 0.1387969067040831,
      "step": 450,
      "vision_kd_loss": 1.2726532906293868,
      "vision_logit_loss": 3.9616181895136835
    },
    {
      "affinity_loss": 0.010629764479235746,
      "emkd_kd_loss": 1.8008181317523122,
      "epoch": 0.0419566482931511,
      "grad_norm": 0.6270830035209656,
      "hard_loss": 1.0824967576352356,
      "joint_kd_loss": 4.1699925177544355,
      "learning_rate": 9.996329635716633e-06,
      "loss": 3.437493896484375,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.1152365502229986,
      "sre_geom_loss": 0.002911870249831736,
      "sre_kd_loss": 6.539166853306815,
      "sre_logit_loss": 6.2649921839754095,
      "sre_span_loss": 0.12858120408374815,
      "step": 500,
      "vision_kd_loss": 1.2431998552381993,
      "vision_logit_loss": 3.9098229756951333
    },
    {
      "affinity_loss": 0.01024358814000152,
      "emkd_kd_loss": 1.781773423179984,
      "epoch": 0.04615231312246621,
      "grad_norm": 0.6800042390823364,
      "hard_loss": 1.0073506324574555,
      "joint_kd_loss": 6.122523047029972,
      "learning_rate": 9.993265686855645e-06,
      "loss": 3.679226989746094,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.1264205457454226,
      "sre_geom_loss": 0.0031517735289543227,
      "sre_kd_loss": 10.463272778056561,
      "sre_logit_loss": 10.18358685581712,
      "sre_span_loss": 0.12209722983650863,
      "step": 550,
      "vision_kd_loss": 1.218563147559762,
      "vision_logit_loss": 3.849893780648708
    },
    {
      "affinity_loss": 0.009414495733799412,
      "emkd_kd_loss": 1.7414812422916293,
      "epoch": 0.05034797795178132,
      "grad_norm": 1.1136304140090942,
      "hard_loss": 1.0071562306169652,
      "joint_kd_loss": 6.343007229790092,
      "learning_rate": 9.989279799224622e-06,
      "loss": 2.9053961181640626,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.1667364392319906,
      "sre_geom_loss": 0.002782820215252286,
      "sre_kd_loss": 10.944533176450058,
      "sre_logit_loss": 10.690765257228632,
      "sre_span_loss": 0.11462696403264999,
      "step": 600,
      "vision_kd_loss": 1.1581130220741034,
      "vision_logit_loss": 3.6910025152564048
    },
    {
      "affinity_loss": 0.009327246435568669,
      "emkd_kd_loss": 1.7406424571946264,
      "epoch": 0.054543642781096434,
      "grad_norm": 0.3908172845840454,
      "hard_loss": 0.9826705935425707,
      "joint_kd_loss": 5.693254618346691,
      "learning_rate": 9.98437270876365e-06,
      "loss": 3.22892333984375,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.1435555524143455,
      "sre_geom_loss": 0.0027936708949152945,
      "sre_kd_loss": 9.645866813389585,
      "sre_logit_loss": 9.397568582647946,
      "sre_span_loss": 0.10861466274131089,
      "step": 650,
      "vision_kd_loss": 1.1688646790385246,
      "vision_logit_loss": 3.742734067738056
    },
    {
      "affinity_loss": 0.008856521506095305,
      "emkd_kd_loss": 1.6962639939412474,
      "epoch": 0.058739307610411545,
      "grad_norm": 0.26912662386894226,
      "hard_loss": 1.0213986446471972,
      "joint_kd_loss": 5.54166726320982,
      "learning_rate": 9.978545321500406e-06,
      "loss": 2.807312316894531,
      "matched_vision_samples": 0.94125,
      "response_kd_loss": 1.1248049491923666,
      "sre_geom_loss": 0.0028460533796948084,
      "sre_kd_loss": 9.387070594914258,
      "sre_logit_loss": 9.138851880927104,
      "sre_span_loss": 0.10591610009782017,
      "step": 700,
      "vision_kd_loss": 1.133861520960927,
      "vision_logit_loss": 3.6497939345240593
    },
    {
      "affinity_loss": 0.009186202498385683,
      "emkd_kd_loss": 1.6823201537132264,
      "epoch": 0.06293497243972665,
      "grad_norm": 0.732519268989563,
      "hard_loss": 0.9304552947553748,
      "joint_kd_loss": 4.474800287783146,
      "learning_rate": 9.971798713382896e-06,
      "loss": 3.08843505859375,
      "matched_vision_samples": 0.96125,
      "response_kd_loss": 1.0091528347162784,
      "sre_geom_loss": 0.0028107746374363353,
      "sre_kd_loss": 7.267280429750681,
      "sre_logit_loss": 7.025190659933724,
      "sre_span_loss": 0.10155105735175311,
      "step": 750,
      "vision_kd_loss": 1.1777437388151883,
      "vision_logit_loss": 3.7923547089099885
    },
    {
      "affinity_loss": 0.008510736097232439,
      "emkd_kd_loss": 1.6778607424907386,
      "epoch": 0.06713063726904177,
      "grad_norm": 0.6371744275093079,
      "hard_loss": 0.9525338502045634,
      "joint_kd_loss": 5.150628855898976,
      "learning_rate": 9.964134130080784e-06,
      "loss": 3.2374188232421877,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.104778724114849,
      "sre_geom_loss": 0.0028105446365987064,
      "sre_kd_loss": 8.623396896217018,
      "sre_logit_loss": 8.383397229674738,
      "sre_span_loss": 0.09947237653192133,
      "step": 800,
      "vision_kd_loss": 1.1254713792353868,
      "vision_logit_loss": 3.6508118963241576
    },
    {
      "affinity_loss": 0.008531417618505656,
      "emkd_kd_loss": 1.6577512727305292,
      "epoch": 0.07132630209835687,
      "grad_norm": 0.8766275644302368,
      "hard_loss": 0.9804236457200023,
      "joint_kd_loss": 4.955442317500711,
      "learning_rate": 9.9555529867554e-06,
      "loss": 3.091214904785156,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0595293612908154,
      "sre_geom_loss": 0.002878225299418773,
      "sre_kd_loss": 8.253133235620334,
      "sre_logit_loss": 8.011497413414764,
      "sre_span_loss": 0.09772449993994087,
      "step": 850,
      "vision_kd_loss": 1.1279865892976522,
      "vision_logit_loss": 3.658804596364498
    },
    {
      "affinity_loss": 0.008224022543872707,
      "emkd_kd_loss": 1.603170555047691,
      "epoch": 0.07552196692767198,
      "grad_norm": 1.1192320585250854,
      "hard_loss": 0.9343678493259722,
      "joint_kd_loss": 5.741752364784479,
      "learning_rate": 9.946056867798442e-06,
      "loss": 3.315203857421875,
      "matched_vision_samples": 0.94375,
      "response_kd_loss": 0.97430172313615,
      "sre_geom_loss": 0.0024983699047004394,
      "sre_kd_loss": 9.88033410041593,
      "sre_logit_loss": 9.661788581162691,
      "sre_span_loss": 0.09362695248331875,
      "step": 900,
      "vision_kd_loss": 1.1160196923464536,
      "vision_logit_loss": 3.641676508784294
    },
    {
      "affinity_loss": 0.008282737977569923,
      "emkd_kd_loss": 1.6720475651323796,
      "epoch": 0.0797176317569871,
      "grad_norm": 0.5376790165901184,
      "hard_loss": 0.8965006697203535,
      "joint_kd_loss": 4.630079071819782,
      "learning_rate": 9.935647526539452e-06,
      "loss": 2.99095703125,
      "matched_vision_samples": 0.9675,
      "response_kd_loss": 1.065589906857167,
      "sre_geom_loss": 0.002624198403714502,
      "sre_kd_loss": 7.588110561873764,
      "sre_logit_loss": 7.366751508832676,
      "sre_span_loss": 0.09014905279502272,
      "step": 950,
      "vision_kd_loss": 1.139252610653639,
      "vision_logit_loss": 3.7287366500496866
    },
    {
      "affinity_loss": 0.008267849877593108,
      "emkd_kd_loss": 1.7091977199539543,
      "epoch": 0.0839132965863022,
      "grad_norm": 0.5611258745193481,
      "hard_loss": 0.8963363857293735,
      "joint_kd_loss": 6.330306726172567,
      "learning_rate": 9.924326884922081e-06,
      "loss": 3.292178955078125,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.1515931018693573,
      "sre_geom_loss": 0.002826799793713519,
      "sre_kd_loss": 10.951415753392503,
      "sre_logit_loss": 10.717405581078491,
      "sre_span_loss": 0.09267009444069117,
      "step": 1000,
      "vision_kd_loss": 1.1334011702239513,
      "vision_logit_loss": 3.7068196880817412
    },
    {
      "affinity_loss": 0.0078417377313599,
      "emkd_kd_loss": 1.6794185964949429,
      "epoch": 0.08810896141561732,
      "grad_norm": 0.24948275089263916,
      "hard_loss": 0.9529905165210948,
      "joint_kd_loss": 5.395914648659527,
      "learning_rate": 9.912097033149227e-06,
      "loss": 3.1896514892578125,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.1453272056801598,
      "sre_geom_loss": 0.002657343666222971,
      "sre_kd_loss": 9.112410765094683,
      "sre_logit_loss": 8.890333531358046,
      "sre_span_loss": 0.08921019102446735,
      "step": 1050,
      "vision_kd_loss": 1.1067549931257963,
      "vision_logit_loss": 3.642846195101738
    },
    {
      "affinity_loss": 0.00793057501607109,
      "emkd_kd_loss": 1.6683725814381614,
      "epoch": 0.09230462624493242,
      "grad_norm": 0.5770359635353088,
      "hard_loss": 0.9368661071019233,
      "joint_kd_loss": 4.969917868636548,
      "learning_rate": 9.898960229297108e-06,
      "loss": 2.9685870361328126,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.1194543396695962,
      "sre_geom_loss": 0.0028399841707090444,
      "sre_kd_loss": 8.271463235970586,
      "sre_logit_loss": 8.04318804197712,
      "sre_span_loss": 0.08627594874240457,
      "step": 1100,
      "vision_kd_loss": 1.1086454091221094,
      "vision_logit_loss": 3.641524138152599
    },
    {
      "affinity_loss": 0.007747870520688593,
      "emkd_kd_loss": 1.6740462223440409,
      "epoch": 0.09650029107424754,
      "grad_norm": 1.0582334995269775,
      "hard_loss": 0.9132483626943576,
      "joint_kd_loss": 4.45749240256846,
      "learning_rate": 9.884918898898338e-06,
      "loss": 3.258453369140625,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.1506317723221582,
      "sre_geom_loss": 0.002749652986398843,
      "sre_kd_loss": 7.240938543463126,
      "sre_logit_loss": 7.017475687439437,
      "sre_span_loss": 0.0859801826858893,
      "step": 1150,
      "vision_kd_loss": 1.0987303363531828,
      "vision_logit_loss": 3.620134293437004
    },
    {
      "affinity_loss": 0.007460454501560889,
      "emkd_kd_loss": 1.5436331636831164,
      "epoch": 0.10069595590356265,
      "grad_norm": 0.47788071632385254,
      "hard_loss": 0.9442737372854754,
      "joint_kd_loss": 4.855623024590313,
      "learning_rate": 9.869975634494094e-06,
      "loss": 3.2839666748046876,
      "matched_vision_samples": 0.93625,
      "response_kd_loss": 0.943022204081085,
      "sre_geom_loss": 0.0025275750143012488,
      "sre_kd_loss": 8.167612950634211,
      "sre_logit_loss": 7.956025611414807,
      "sre_span_loss": 0.0852085403772071,
      "step": 1200,
      "vision_kd_loss": 1.0721220614761113,
      "vision_logit_loss": 3.542442802786827
    },
    {
      "affinity_loss": 0.0077990044985199345,
      "emkd_kd_loss": 1.6897247751057147,
      "epoch": 0.10489162073287775,
      "grad_norm": 0.3269062936306,
      "hard_loss": 0.8588140244941769,
      "joint_kd_loss": 3.464735939502716,
      "learning_rate": 9.854133195155422e-06,
      "loss": 2.7104766845703123,
      "matched_vision_samples": 0.96125,
      "response_kd_loss": 1.1389029961004327,
      "sre_geom_loss": 0.002456971306429523,
      "sre_kd_loss": 5.239747077096253,
      "sre_logit_loss": 5.035102699586423,
      "sre_span_loss": 0.08179586586542427,
      "step": 1250,
      "vision_kd_loss": 1.120273275822401,
      "vision_logit_loss": 3.7011926513910294
    },
    {
      "affinity_loss": 0.007546176708128769,
      "emkd_kd_loss": 1.593918388299644,
      "epoch": 0.10908728556219287,
      "grad_norm": 0.3789830803871155,
      "hard_loss": 0.9449307548270736,
      "joint_kd_loss": 5.538269223719835,
      "learning_rate": 9.83739450597383e-06,
      "loss": 3.0501153564453123,
      "matched_vision_samples": 0.94375,
      "response_kd_loss": 0.9940025572794638,
      "sre_geom_loss": 0.002503233913702161,
      "sre_kd_loss": 9.482620102893561,
      "sre_logit_loss": 9.274204113173182,
      "sre_span_loss": 0.08325427509378641,
      "step": 1300,
      "vision_kd_loss": 1.0969171100109816,
      "vision_logit_loss": 3.6330507624149324
    },
    {
      "affinity_loss": 0.007758410104725044,
      "emkd_kd_loss": 1.629450579583645,
      "epoch": 0.11328295039150797,
      "grad_norm": 0.22131521999835968,
      "hard_loss": 0.9212364768793486,
      "joint_kd_loss": 5.588119214400649,
      "learning_rate": 9.819762657521195e-06,
      "loss": 3.20559814453125,
      "matched_vision_samples": 0.96,
      "response_kd_loss": 1.0308997915598228,
      "sre_geom_loss": 0.0024585032149689744,
      "sre_kd_loss": 9.546787825236096,
      "sre_logit_loss": 9.341000518881483,
      "sre_span_loss": 0.082862205686979,
      "step": 1350,
      "vision_kd_loss": 1.1140006835013627,
      "vision_logit_loss": 3.680161726474762
    },
    {
      "affinity_loss": 0.007353854610410053,
      "emkd_kd_loss": 1.6019624833017587,
      "epoch": 0.11747861522082309,
      "grad_norm": 0.3280599117279053,
      "hard_loss": 0.9004752969659603,
      "joint_kd_loss": 3.6806995622068643,
      "learning_rate": 9.801240905279139e-06,
      "loss": 2.8787811279296873,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.0164783956027759,
      "sre_geom_loss": 0.0023750548156175454,
      "sre_kd_loss": 5.759436676697805,
      "sre_logit_loss": 5.562194854167756,
      "sre_span_loss": 0.07848905692342668,
      "step": 1400,
      "vision_kd_loss": 1.0937232840061188,
      "vision_logit_loss": 3.639507673382759
    },
    {
      "affinity_loss": 0.007493529361672699,
      "emkd_kd_loss": 1.5972844458743929,
      "epoch": 0.1216742800501382,
      "grad_norm": 1.1355030536651611,
      "hard_loss": 0.892752265702402,
      "joint_kd_loss": 4.734732951670885,
      "learning_rate": 9.781832669037944e-06,
      "loss": 3.2413323974609374,
      "matched_vision_samples": 0.96375,
      "response_kd_loss": 0.9930345976363742,
      "sre_geom_loss": 0.002463084682468306,
      "sre_kd_loss": 7.8721813977696,
      "sre_logit_loss": 7.667632728292374,
      "sre_span_loss": 0.08139453026000411,
      "step": 1450,
      "vision_kd_loss": 1.100767144933343,
      "vision_logit_loss": 3.653715642392635
    },
    {
      "affinity_loss": 0.007262250056082848,
      "emkd_kd_loss": 1.600868139564991,
      "epoch": 0.1258699448794533,
      "grad_norm": 0.3198093771934509,
      "hard_loss": 0.9210896198291448,
      "joint_kd_loss": 4.769744224213063,
      "learning_rate": 9.761541532265143e-06,
      "loss": 3.143026123046875,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0480320521937119,
      "sre_geom_loss": 0.00267595303142528,
      "sre_kd_loss": 7.938620240008459,
      "sre_logit_loss": 7.723052358608693,
      "sre_span_loss": 0.081770222466439,
      "step": 1500,
      "vision_kd_loss": 1.07685211494565,
      "vision_logit_loss": 3.5811834481358527
    },
    {
      "affinity_loss": 0.007438291998696513,
      "emkd_kd_loss": 1.6051473029330372,
      "epoch": 0.1300656097087684,
      "grad_norm": 0.25582996010780334,
      "hard_loss": 0.8924661307270071,
      "joint_kd_loss": 5.9693590053170915,
      "learning_rate": 9.740371241443865e-06,
      "loss": 2.8792425537109376,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0446071859425274,
      "sre_geom_loss": 0.0028343278783864887,
      "sre_kd_loss": 10.333570626443253,
      "sre_logit_loss": 10.108412887116428,
      "sre_span_loss": 0.08344129939563573,
      "step": 1550,
      "vision_kd_loss": 1.082843709141016,
      "vision_logit_loss": 3.5875456327199937
    },
    {
      "affinity_loss": 0.0072354324845946396,
      "emkd_kd_loss": 1.5957241693511606,
      "epoch": 0.13426127453808354,
      "grad_norm": 0.1896904706954956,
      "hard_loss": 0.9392215549735556,
      "joint_kd_loss": 5.320351254530251,
      "learning_rate": 9.718325705381115e-06,
      "loss": 3.1695620727539064,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.0300114394591993,
      "sre_geom_loss": 0.002539037717319985,
      "sre_kd_loss": 9.04497830583714,
      "sre_logit_loss": 8.83714560182765,
      "sre_span_loss": 0.08088083814596757,
      "step": 1600,
      "vision_kd_loss": 1.0807184502482414,
      "vision_logit_loss": 3.599330554306507
    },
    {
      "affinity_loss": 0.007399906954087782,
      "emkd_kd_loss": 1.6256348850764335,
      "epoch": 0.13845693936739864,
      "grad_norm": 0.6293073296546936,
      "hard_loss": 0.9477892382144637,
      "joint_kd_loss": 5.249184664264321,
      "learning_rate": 9.695408994486054e-06,
      "loss": 3.1134829711914063,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.061127561877629,
      "sre_geom_loss": 0.002759405155120476,
      "sre_kd_loss": 8.872734487229026,
      "sre_logit_loss": 8.655035234730457,
      "sre_span_loss": 0.07972901268396526,
      "step": 1650,
      "vision_kd_loss": 1.0950711032003164,
      "vision_logit_loss": 3.640293719172478
    },
    {
      "affinity_loss": 0.007495863645162899,
      "emkd_kd_loss": 1.6153194416314363,
      "epoch": 0.14265260419671374,
      "grad_norm": 1.4137108325958252,
      "hard_loss": 0.9273266931568287,
      "joint_kd_loss": 6.762068729810417,
      "learning_rate": 9.671625340018462e-06,
      "loss": 3.2444952392578124,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0265003637871997,
      "sre_geom_loss": 0.0024444927851801877,
      "sre_kd_loss": 11.90881804968696,
      "sre_logit_loss": 11.70670425298682,
      "sre_span_loss": 0.07988910383312031,
      "step": 1700,
      "vision_kd_loss": 1.1020692581683398,
      "vision_logit_loss": 3.658690667748451
    },
    {
      "affinity_loss": 0.007010118253238034,
      "emkd_kd_loss": 1.5459245995804667,
      "epoch": 0.14684826902602885,
      "grad_norm": 0.2572149336338043,
      "hard_loss": 0.9883559027370211,
      "joint_kd_loss": 6.078529685102403,
      "learning_rate": 9.64697913330748e-06,
      "loss": 3.2147555541992188,
      "matched_vision_samples": 0.93375,
      "response_kd_loss": 1.0005990732174632,
      "sre_geom_loss": 0.002500235606446992,
      "sre_kd_loss": 10.611134763387962,
      "sre_logit_loss": 10.406769610082264,
      "sre_span_loss": 0.07935340851778165,
      "step": 1750,
      "vision_kd_loss": 1.0456250592321157,
      "vision_logit_loss": 3.4814884144067766
    },
    {
      "affinity_loss": 0.007161422515055165,
      "emkd_kd_loss": 1.603824936002493,
      "epoch": 0.15104393385534395,
      "grad_norm": 0.22376351058483124,
      "hard_loss": 0.952542966563451,
      "joint_kd_loss": 4.430673902705312,
      "learning_rate": 9.621474924940829e-06,
      "loss": 3.213284912109375,
      "matched_vision_samples": 0.96125,
      "response_kd_loss": 1.0402766009109745,
      "sre_geom_loss": 0.0022953954891324815,
      "sre_kd_loss": 7.257522879433818,
      "sre_logit_loss": 7.065641113355523,
      "sre_span_loss": 0.07711190198082477,
      "step": 1800,
      "vision_kd_loss": 1.0836866325885057,
      "vision_logit_loss": 3.61860427737236
    },
    {
      "affinity_loss": 0.007276012504880782,
      "emkd_kd_loss": 1.6375696281716228,
      "epoch": 0.15523959868465909,
      "grad_norm": 0.9410597085952759,
      "hard_loss": 0.918941433377513,
      "joint_kd_loss": 4.9934254097566,
      "learning_rate": 9.59511742392459e-06,
      "loss": 3.24320556640625,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.1178397474967006,
      "sre_geom_loss": 0.0025033994906539193,
      "sre_kd_loss": 8.349281168146991,
      "sre_logit_loss": 8.146673047953518,
      "sre_span_loss": 0.07743806984974072,
      "step": 1850,
      "vision_kd_loss": 1.0786497522890568,
      "vision_logit_loss": 3.5869977596402167
    },
    {
      "affinity_loss": 0.007047707361343782,
      "emkd_kd_loss": 1.6659263388440013,
      "epoch": 0.1594352635139742,
      "grad_norm": 0.9024304151535034,
      "hard_loss": 0.8920237209226434,
      "joint_kd_loss": 4.224776862449944,
      "learning_rate": 9.56791149681376e-06,
      "loss": 2.8613201904296877,
      "matched_vision_samples": 0.96125,
      "response_kd_loss": 1.181312774029211,
      "sre_geom_loss": 0.002728360499546625,
      "sre_kd_loss": 6.7836274111829695,
      "sre_logit_loss": 6.5715536191908175,
      "sre_span_loss": 0.07565575405256823,
      "step": 1900,
      "vision_kd_loss": 1.0752699528634548,
      "vision_logit_loss": 3.5963090753555296
    },
    {
      "affinity_loss": 0.0071286970193614255,
      "emkd_kd_loss": 1.6099900530278684,
      "epoch": 0.1636309283432893,
      "grad_norm": 0.8334711194038391,
      "hard_loss": 0.9439037961174472,
      "joint_kd_loss": 4.751788458563388,
      "learning_rate": 9.539862166813699e-06,
      "loss": 3.335954284667969,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0744904914919788,
      "sre_geom_loss": 0.002454849294235828,
      "sre_kd_loss": 7.8935868234233935,
      "sre_logit_loss": 7.69257837297977,
      "sre_span_loss": 0.07826603478053584,
      "step": 1950,
      "vision_kd_loss": 1.0727448086440563,
      "vision_logit_loss": 3.5781095358729362
    },
    {
      "affinity_loss": 0.00704633162211394,
      "emkd_kd_loss": 1.5617236853204668,
      "epoch": 0.1678265931726044,
      "grad_norm": 0.28601643443107605,
      "hard_loss": 0.8854854521634116,
      "joint_kd_loss": 5.842119127810001,
      "learning_rate": 9.510974612852672e-06,
      "loss": 3.2514361572265624,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 0.9716967742049973,
      "sre_geom_loss": 0.002170386377716227,
      "sre_kd_loss": 10.12251453964971,
      "sre_logit_loss": 9.939188662971137,
      "sre_span_loss": 0.07480658424785361,
      "step": 2000,
      "vision_kd_loss": 1.0758752996474505,
      "vision_logit_loss": 3.5988680297136306
    },
    {
      "affinity_loss": 0.007066695174144115,
      "emkd_kd_loss": 1.5975545309111476,
      "epoch": 0.1720222580019195,
      "grad_norm": 0.43123486638069153,
      "hard_loss": 0.9311993248548243,
      "joint_kd_loss": 6.890781584493816,
      "learning_rate": 9.48125416862563e-06,
      "loss": 3.0018017578125,
      "matched_vision_samples": 0.945,
      "response_kd_loss": 1.036269549840763,
      "sre_geom_loss": 0.002408831078909088,
      "sre_kd_loss": 12.184008594593033,
      "sre_logit_loss": 11.98697442390956,
      "sre_span_loss": 0.07659267783164977,
      "step": 2050,
      "vision_kd_loss": 1.0794197565317154,
      "vision_logit_loss": 3.611009507775307
    },
    {
      "affinity_loss": 0.006823928747035097,
      "emkd_kd_loss": 1.5344908707588911,
      "epoch": 0.17621792283123464,
      "grad_norm": 0.4727615416049957,
      "hard_loss": 0.9015850974950809,
      "joint_kd_loss": 6.445970040000975,
      "learning_rate": 9.450706321609407e-06,
      "loss": 3.1042379760742187,
      "matched_vision_samples": 0.9425,
      "response_kd_loss": 0.9656801535843852,
      "sre_geom_loss": 0.0023534183070218263,
      "sre_kd_loss": 11.35744913706556,
      "sre_logit_loss": 11.166293565289234,
      "sre_span_loss": 0.07348472610348836,
      "step": 2100,
      "vision_kd_loss": 1.0516507943719626,
      "vision_logit_loss": 3.52421030074358
    },
    {
      "affinity_loss": 0.006965450907009654,
      "emkd_kd_loss": 1.5811904757097364,
      "epoch": 0.18041358766054974,
      "grad_norm": 0.32083404064178467,
      "hard_loss": 0.9010861140772249,
      "joint_kd_loss": 5.316080280505121,
      "learning_rate": 9.419336712049543e-06,
      "loss": 3.063535461425781,
      "matched_vision_samples": 0.945,
      "response_kd_loss": 1.0436781407143463,
      "sre_geom_loss": 0.0022162318059119456,
      "sre_kd_loss": 9.05097004535608,
      "sre_logit_loss": 8.865467267911882,
      "sre_span_loss": 0.07469116407446563,
      "step": 2150,
      "vision_kd_loss": 1.0593514043092727,
      "vision_logit_loss": 3.5408605280518533
    },
    {
      "affinity_loss": 0.006897779165010434,
      "emkd_kd_loss": 1.5390548553317784,
      "epoch": 0.18460925248986484,
      "grad_norm": 0.45092564821243286,
      "hard_loss": 0.9526381313449565,
      "joint_kd_loss": 6.343198441378772,
      "learning_rate": 9.387151131918876e-06,
      "loss": 3.26021728515625,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 0.9742361437530098,
      "sre_geom_loss": 0.002071611509429303,
      "sre_kd_loss": 11.147341927504167,
      "sre_logit_loss": 10.968940219696961,
      "sre_span_loss": 0.07482110804179683,
      "step": 2200,
      "vision_kd_loss": 1.0519367834180593,
      "vision_logit_loss": 3.5179692208766937
    },
    {
      "affinity_loss": 0.007112885545066092,
      "emkd_kd_loss": 1.5882415500842035,
      "epoch": 0.18880491731917995,
      "grad_norm": 1.0674676895141602,
      "hard_loss": 0.9460364071693038,
      "joint_kd_loss": 5.236769439429045,
      "learning_rate": 9.35415552384815e-06,
      "loss": 3.32883544921875,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.022132383557041,
      "sre_geom_loss": 0.0023793898341507714,
      "sre_kd_loss": 8.885297321802936,
      "sre_logit_loss": 8.693172140397946,
      "sre_span_loss": 0.07315571539336815,
      "step": 2250,
      "vision_kd_loss": 1.0771753576397896,
      "vision_logit_loss": 3.597412869334221
    },
    {
      "affinity_loss": 0.00706774150574347,
      "emkd_kd_loss": 1.601085498072207,
      "epoch": 0.19300058214849508,
      "grad_norm": 0.2427961379289627,
      "hard_loss": 0.944849043117356,
      "joint_kd_loss": 5.55017225548625,
      "learning_rate": 9.320355980028768e-06,
      "loss": 3.1178240966796875,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.0568323287488237,
      "sre_geom_loss": 0.0024467601280281316,
      "sre_kd_loss": 9.499259040621109,
      "sre_logit_loss": 9.30308837611694,
      "sre_span_loss": 0.07383264386327937,
      "step": 2300,
      "vision_kd_loss": 1.0726693327724934,
      "vision_logit_loss": 3.583903174102306
    },
    {
      "affinity_loss": 0.006826247753051576,
      "emkd_kd_loss": 1.581934796832502,
      "epoch": 0.19719624697781019,
      "grad_norm": 0.20863142609596252,
      "hard_loss": 0.9331530989783369,
      "joint_kd_loss": 6.023809741549194,
      "learning_rate": 9.285758741087976e-06,
      "loss": 3.146928405761719,
      "matched_vision_samples": 0.94125,
      "response_kd_loss": 1.0743424027868924,
      "sre_geom_loss": 0.0024583219147788784,
      "sre_kd_loss": 10.465684685721062,
      "sre_logit_loss": 10.269166911931825,
      "sre_span_loss": 0.07360173858702183,
      "step": 2350,
      "vision_kd_loss": 1.0447635950893164,
      "vision_logit_loss": 3.4964296010136606
    },
    {
      "affinity_loss": 0.006823228530702181,
      "emkd_kd_loss": 1.5358967689797283,
      "epoch": 0.2013919118071253,
      "grad_norm": 0.16194196045398712,
      "hard_loss": 0.8990054182298627,
      "joint_kd_loss": 4.889795694164932,
      "learning_rate": 9.250370194936591e-06,
      "loss": 2.9036224365234373,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 0.9728520753990597,
      "sre_geom_loss": 0.002209656018460464,
      "sre_kd_loss": 8.243694638595917,
      "sre_logit_loss": 8.061223346181214,
      "sre_span_loss": 0.07198854852933437,
      "step": 2400,
      "vision_kd_loss": 1.0494707329571247,
      "vision_logit_loss": 3.515560081601143
    },
    {
      "affinity_loss": 0.00691791719320463,
      "emkd_kd_loss": 1.6042517692036926,
      "epoch": 0.2055875766364404,
      "grad_norm": 0.3785500228404999,
      "hard_loss": 0.9047605347468197,
      "joint_kd_loss": 5.971395424641669,
      "learning_rate": 9.214196875589577e-06,
      "loss": 3.293746032714844,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0774898055927407,
      "sre_geom_loss": 0.002441201095967216,
      "sre_kd_loss": 10.338538991538808,
      "sre_logit_loss": 10.143074984112172,
      "sre_span_loss": 0.07340390544384717,
      "step": 2450,
      "vision_kd_loss": 1.0655068653076887,
      "vision_logit_loss": 3.5702357417345048
    },
    {
      "affinity_loss": 0.006857141503423918,
      "emkd_kd_loss": 1.6005623802542686,
      "epoch": 0.2097832414657555,
      "grad_norm": 0.7503357529640198,
      "hard_loss": 0.9121618518705873,
      "joint_kd_loss": 6.668967695012689,
      "learning_rate": 9.17724546195962e-06,
      "loss": 3.526270751953125,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 1.0794904356538426,
      "sre_geom_loss": 0.002671196049271032,
      "sre_kd_loss": 11.737372998115607,
      "sre_logit_loss": 11.531220142911188,
      "sre_span_loss": 0.07259306025458499,
      "step": 2500,
      "vision_kd_loss": 1.06081716299057,
      "vision_logit_loss": 3.557554500699043
    },
    {
      "affinity_loss": 0.006789842607104219,
      "emkd_kd_loss": 1.5575766525417567,
      "epoch": 0.21397890629507063,
      "grad_norm": 0.6593822240829468,
      "hard_loss": 0.8923383242687851,
      "joint_kd_loss": 6.217473382093012,
      "learning_rate": 9.139522776623963e-06,
      "loss": 3.0881121826171873,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 0.9966745441980811,
      "sre_geom_loss": 0.002196180973362516,
      "sre_kd_loss": 10.877370015271008,
      "sre_logit_loss": 10.696178532725899,
      "sre_span_loss": 0.0713824265357107,
      "step": 2550,
      "vision_kd_loss": 1.059239379838109,
      "vision_logit_loss": 3.5579732662439345
    },
    {
      "affinity_loss": 0.006756929741823114,
      "emkd_kd_loss": 1.569112147167325,
      "epoch": 0.21817457112438574,
      "grad_norm": 0.7233375310897827,
      "hard_loss": 0.8969858224088239,
      "joint_kd_loss": 3.9150295232236387,
      "learning_rate": 9.101035784564703e-06,
      "loss": 2.756353759765625,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0143741392399261,
      "sre_geom_loss": 0.0023375144374455205,
      "sre_kd_loss": 6.260946966735646,
      "sre_logit_loss": 6.072581339660392,
      "sre_span_loss": 0.07148998569929972,
      "step": 2600,
      "vision_kd_loss": 1.061925078779459,
      "vision_logit_loss": 3.5720073392987253
    },
    {
      "affinity_loss": 0.006940552941523492,
      "emkd_kd_loss": 1.5933731453865767,
      "epoch": 0.22237023595370084,
      "grad_norm": 0.2707875669002533,
      "hard_loss": 0.9031434797750262,
      "joint_kd_loss": 5.64374238833785,
      "learning_rate": 9.061791591882807e-06,
      "loss": 3.30961669921875,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.0386717424691505,
      "sre_geom_loss": 0.002366062043400978,
      "sre_kd_loss": 9.694111701496876,
      "sre_logit_loss": 9.504354472397827,
      "sre_span_loss": 0.07145405851770192,
      "step": 2650,
      "vision_kd_loss": 1.0740372755378484,
      "vision_logit_loss": 3.6020938104391096
    },
    {
      "affinity_loss": 0.006881514233828057,
      "emkd_kd_loss": 1.5750772021710873,
      "epoch": 0.22656590078301594,
      "grad_norm": 0.8134822249412537,
      "hard_loss": 0.8625151714464664,
      "joint_kd_loss": 5.400984864346683,
      "learning_rate": 9.021797444486064e-06,
      "loss": 3.3001361083984375,
      "matched_vision_samples": 0.96,
      "response_kd_loss": 0.9942540359715712,
      "sre_geom_loss": 0.002248154998308394,
      "sre_kd_loss": 9.226892551793718,
      "sre_logit_loss": 9.043261147728481,
      "sre_span_loss": 0.07122363924281672,
      "step": 2700,
      "vision_kd_loss": 1.0779501812160015,
      "vision_logit_loss": 3.623649302124977
    },
    {
      "affinity_loss": 0.00669571884238394,
      "emkd_kd_loss": 1.5259974868223072,
      "epoch": 0.23076156561233105,
      "grad_norm": 0.3058062195777893,
      "hard_loss": 0.9293799785198644,
      "joint_kd_loss": 3.382855116575956,
      "learning_rate": 8.981060726751225e-06,
      "loss": 2.638860778808594,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 0.9418539664837136,
      "sre_geom_loss": 0.002066673666950081,
      "sre_kd_loss": 5.239712700000964,
      "sre_logit_loss": 5.068446019184775,
      "sre_span_loss": 0.0679329722514376,
      "step": 2750,
      "vision_kd_loss": 1.0550705007463694,
      "vision_logit_loss": 3.5507101091742514
    },
    {
      "affinity_loss": 0.006732612085179426,
      "emkd_kd_loss": 1.5295603031106293,
      "epoch": 0.23495723044164618,
      "grad_norm": 0.5565466284751892,
      "hard_loss": 0.8891497871030151,
      "joint_kd_loss": 4.506467787437141,
      "learning_rate": 8.939588960160583e-06,
      "loss": 2.744828796386719,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 0.969845543607953,
      "sre_geom_loss": 0.0022482199482485045,
      "sre_kd_loss": 7.48337524454575,
      "sre_logit_loss": 7.299943983296398,
      "sre_span_loss": 0.07102022263687104,
      "step": 2800,
      "vision_kd_loss": 1.0446375305950641,
      "vision_logit_loss": 3.505288910269737
    },
    {
      "affinity_loss": 0.006885603545524646,
      "emkd_kd_loss": 1.5693590710312129,
      "epoch": 0.23915289527096129,
      "grad_norm": 0.5540454983711243,
      "hard_loss": 0.9182299824732764,
      "joint_kd_loss": 3.448753966987133,
      "learning_rate": 8.897389801913232e-06,
      "loss": 2.851566162109375,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.0137872934474945,
      "sre_geom_loss": 0.0026025321311681183,
      "sre_kd_loss": 5.328148858365603,
      "sre_logit_loss": 5.126732913111919,
      "sre_span_loss": 0.07128929927246645,
      "step": 2850,
      "vision_kd_loss": 1.0624654249101877,
      "vision_logit_loss": 3.561301344037056
    },
    {
      "affinity_loss": 0.006652447803935502,
      "emkd_kd_loss": 1.5585881155729293,
      "epoch": 0.2433485601002764,
      "grad_norm": 0.5124243497848511,
      "hard_loss": 0.8629311413632058,
      "joint_kd_loss": 6.369412845186889,
      "learning_rate": 8.854471043511261e-06,
      "loss": 3.2918670654296873,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0144607731628092,
      "sre_geom_loss": 0.0020727676829640274,
      "sre_kd_loss": 11.180237593338825,
      "sre_logit_loss": 11.007175055155821,
      "sre_span_loss": 0.0694241470634006,
      "step": 2900,
      "vision_kd_loss": 1.051357729509473,
      "vision_logit_loss": 3.5401861396431924
    },
    {
      "affinity_loss": 0.0067424417354050095,
      "emkd_kd_loss": 1.5468588504195213,
      "epoch": 0.2475442249295915,
      "grad_norm": 0.3962687849998474,
      "hard_loss": 0.908165611883287,
      "joint_kd_loss": 5.754400056600571,
      "learning_rate": 8.810840609321174e-06,
      "loss": 3.0675860595703126,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 0.9887157297397289,
      "sre_geom_loss": 0.0020535555517432157,
      "sre_kd_loss": 9.961941224550829,
      "sre_logit_loss": 9.789051118246862,
      "sre_span_loss": 0.0702123937336728,
      "step": 2950,
      "vision_kd_loss": 1.0525009868294,
      "vision_logit_loss": 3.5357597789168356
    },
    {
      "affinity_loss": 0.006754972654161975,
      "emkd_kd_loss": 1.534355593472719,
      "epoch": 0.2517398897589066,
      "grad_norm": 0.3098095953464508,
      "hard_loss": 0.8909413302350732,
      "joint_kd_loss": 4.884705157801509,
      "learning_rate": 8.766506555110746e-06,
      "loss": 2.93429931640625,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 0.9478493429585114,
      "sre_geom_loss": 0.0020537026286729086,
      "sre_kd_loss": 8.235054750246928,
      "sre_logit_loss": 8.063531488929875,
      "sre_span_loss": 0.06883812804706395,
      "step": 3000,
      "vision_kd_loss": 1.060430922806263,
      "vision_logit_loss": 3.5662264242768287
    },
    {
      "affinity_loss": 0.006747255748778116,
      "emkd_kd_loss": 1.5563178339973092,
      "epoch": 0.2559355545882217,
      "grad_norm": 0.24828097224235535,
      "hard_loss": 0.9156953575718034,
      "joint_kd_loss": 5.229247620254755,
      "learning_rate": 8.72147706656165e-06,
      "loss": 2.87118408203125,
      "matched_vision_samples": 0.96375,
      "response_kd_loss": 0.9865488301229789,
      "sre_geom_loss": 0.002299715497732677,
      "sre_kd_loss": 8.902177372863516,
      "sre_logit_loss": 8.71869135325891,
      "sre_span_loss": 0.06850030892994255,
      "step": 3050,
      "vision_kd_loss": 1.0630434165149927,
      "vision_logit_loss": 3.5774480855464934
    },
    {
      "affinity_loss": 0.006595587868941948,
      "emkd_kd_loss": 1.6016887520253658,
      "epoch": 0.2601312194175368,
      "grad_norm": 0.5095158815383911,
      "hard_loss": 0.902862676013574,
      "joint_kd_loss": 5.32996002137661,
      "learning_rate": 8.675760457758072e-06,
      "loss": 2.935427551269531,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0984545702313904,
      "sre_geom_loss": 0.002190055581961303,
      "sre_kd_loss": 9.058231358174234,
      "sre_logit_loss": 8.88122906043136,
      "sre_span_loss": 0.06749951171455905,
      "step": 3100,
      "vision_kd_loss": 1.0524614672362804,
      "vision_logit_loss": 3.550287082195282
    },
    {
      "affinity_loss": 0.006622081481909845,
      "emkd_kd_loss": 1.552650042064488,
      "epoch": 0.26432688424685197,
      "grad_norm": 1.3601962327957153,
      "hard_loss": 0.9375182107513319,
      "joint_kd_loss": 6.099188229367137,
      "learning_rate": 8.629365169651633e-06,
      "loss": 2.935041809082031,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.001301818095135,
      "sre_geom_loss": 0.0022774058333951074,
      "sre_kd_loss": 10.645726469587535,
      "sre_logit_loss": 10.463798063404683,
      "sre_span_loss": 0.06805811589350924,
      "step": 3150,
      "vision_kd_loss": 1.0519991332292558,
      "vision_logit_loss": 3.5457883766293525
    },
    {
      "affinity_loss": 0.006561928905139212,
      "emkd_kd_loss": 1.5457255496829747,
      "epoch": 0.26852254907616707,
      "grad_norm": 0.45334234833717346,
      "hard_loss": 0.8894868849565865,
      "joint_kd_loss": 4.902988005876541,
      "learning_rate": 8.58229976850288e-06,
      "loss": 3.0622662353515624,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 0.9937573422207834,
      "sre_geom_loss": 0.0020702532567946717,
      "sre_kd_loss": 8.260250487257727,
      "sre_logit_loss": 8.088480861834832,
      "sre_span_loss": 0.06825692654354498,
      "step": 3200,
      "vision_kd_loss": 1.0488468796014785,
      "vision_logit_loss": 3.5391946268081664
    },
    {
      "affinity_loss": 0.006705335322476458,
      "emkd_kd_loss": 1.5884665996022522,
      "epoch": 0.2727182139054822,
      "grad_norm": 0.17113995552062988,
      "hard_loss": 0.8804279260517797,
      "joint_kd_loss": 4.313549602106214,
      "learning_rate": 8.534572944299641e-06,
      "loss": 2.985376892089844,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0642861297278206,
      "sre_geom_loss": 0.002175462510812736,
      "sre_kd_loss": 7.038632613611408,
      "sre_logit_loss": 6.86089804138086,
      "sre_span_loss": 0.0689614615170285,
      "step": 3250,
      "vision_kd_loss": 1.0563235349208115,
      "vision_logit_loss": 3.554760606586933
    },
    {
      "affinity_loss": 0.0067146188058541155,
      "emkd_kd_loss": 1.5972117917612194,
      "epoch": 0.2769138787347973,
      "grad_norm": 0.4675114154815674,
      "hard_loss": 0.8351456207066803,
      "joint_kd_loss": 4.726926428079605,
      "learning_rate": 8.48619350915254e-06,
      "loss": 3.0922509765625,
      "matched_vision_samples": 0.9425,
      "response_kd_loss": 1.095135441029563,
      "sre_geom_loss": 0.0020944601646702664,
      "sre_kd_loss": 7.856641036798246,
      "sre_logit_loss": 7.684946935858752,
      "sre_span_loss": 0.06697102580917999,
      "step": 3300,
      "vision_kd_loss": 1.0496440722048284,
      "vision_logit_loss": 3.5271144086122512
    },
    {
      "affinity_loss": 0.006655108226113953,
      "emkd_kd_loss": 1.560279670730233,
      "epoch": 0.2811095435641124,
      "grad_norm": 0.17583906650543213,
      "hard_loss": 0.8835599966683367,
      "joint_kd_loss": 3.9512256394326686,
      "learning_rate": 8.437170395667957e-06,
      "loss": 3.0341317749023435,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0047820424925658,
      "sre_geom_loss": 0.0022778711552274445,
      "sre_kd_loss": 6.342171575948596,
      "sre_logit_loss": 6.159711383761897,
      "sre_span_loss": 0.06856666197068989,
      "step": 3350,
      "vision_kd_loss": 1.0578886484354735,
      "vision_logit_loss": 3.5660437700152396
    },
    {
      "affinity_loss": 0.006659650770016015,
      "emkd_kd_loss": 1.5622975553758442,
      "epoch": 0.2853052083934275,
      "grad_norm": 0.3499339818954468,
      "hard_loss": 0.8354579301565537,
      "joint_kd_loss": 3.878128137998283,
      "learning_rate": 8.387512655298746e-06,
      "loss": 2.6673849487304686,
      "matched_vision_samples": 0.9675,
      "response_kd_loss": 1.0040961749440702,
      "sre_geom_loss": 0.002056927529472077,
      "sre_kd_loss": 6.1939586751768365,
      "sre_logit_loss": 6.02426673961425,
      "sre_span_loss": 0.06684551384998486,
      "step": 3400,
      "vision_kd_loss": 1.060249472334981,
      "vision_logit_loss": 3.575032812654972
    },
    {
      "affinity_loss": 0.006618975211749784,
      "emkd_kd_loss": 1.5338535383343697,
      "epoch": 0.2895008732227426,
      "grad_norm": 0.3909642696380615,
      "hard_loss": 0.9641090929335405,
      "joint_kd_loss": 4.4085454539209605,
      "learning_rate": 8.337229456673004e-06,
      "loss": 2.9537640380859376,
      "matched_vision_samples": 0.9625,
      "response_kd_loss": 0.9688721635820684,
      "sre_geom_loss": 0.0019173266685459111,
      "sre_kd_loss": 7.283237373228185,
      "sre_logit_loss": 7.1199940232909285,
      "sre_span_loss": 0.06737703603459522,
      "step": 3450,
      "vision_kd_loss": 1.049417457729578,
      "vision_logit_loss": 3.535772305428982
    },
    {
      "affinity_loss": 0.006561618639680091,
      "emkd_kd_loss": 1.5598104925081133,
      "epoch": 0.2936965380520577,
      "grad_norm": 0.4845067858695984,
      "hard_loss": 0.8640036875697297,
      "joint_kd_loss": 4.293961208835245,
      "learning_rate": 8.286330083901213e-06,
      "loss": 3.0932443237304685,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0239063774630768,
      "sre_geom_loss": 0.002201156283887542,
      "sre_kd_loss": 7.028111910820007,
      "sre_logit_loss": 6.850398038966814,
      "sre_span_loss": 0.06765610361704603,
      "step": 3500,
      "vision_kd_loss": 1.0478573038429022,
      "vision_logit_loss": 3.5352673557400704
    },
    {
      "affinity_loss": 0.0064316627659718505,
      "emkd_kd_loss": 1.5345601219311356,
      "epoch": 0.2978922028813728,
      "grad_norm": 0.42399707436561584,
      "hard_loss": 0.909223635087692,
      "joint_kd_loss": 6.3396897726133465,
      "learning_rate": 8.23482393486205e-06,
      "loss": 3.0902981567382812,
      "matched_vision_samples": 0.9375,
      "response_kd_loss": 1.0033319059980021,
      "sre_geom_loss": 0.0020791236652002,
      "sre_kd_loss": 11.144819452930243,
      "sre_logit_loss": 10.973042759657256,
      "sre_span_loss": 0.06782046176260337,
      "step": 3550,
      "vision_kd_loss": 1.0328941709548234,
      "vision_logit_loss": 3.488410409092903
    },
    {
      "affinity_loss": 0.00660312580439495,
      "emkd_kd_loss": 1.588303063735366,
      "epoch": 0.3020878677106879,
      "grad_norm": 0.5027792453765869,
      "hard_loss": 0.89064754214618,
      "joint_kd_loss": 6.132092996425927,
      "learning_rate": 8.182720519467197e-06,
      "loss": 3.4136093139648436,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.0655244079915906,
      "sre_geom_loss": 0.0022738866777521948,
      "sre_kd_loss": 10.675882884440943,
      "sre_logit_loss": 10.495159238615306,
      "sre_span_loss": 0.06702932040905579,
      "step": 3600,
      "vision_kd_loss": 1.0555408638715744,
      "vision_logit_loss": 3.5618508768081667
    },
    {
      "affinity_loss": 0.006625712813111022,
      "emkd_kd_loss": 1.5751216847077012,
      "epoch": 0.30628353254000307,
      "grad_norm": 0.2979172468185425,
      "hard_loss": 0.9325481648716232,
      "joint_kd_loss": 5.539451133944095,
      "learning_rate": 8.130029457905469e-06,
      "loss": 2.96587890625,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.0692036287127167,
      "sre_geom_loss": 0.0024424593184022483,
      "sre_kd_loss": 9.503780587613583,
      "sre_logit_loss": 9.312001437685685,
      "sre_span_loss": 0.06965617783134803,
      "step": 3650,
      "vision_kd_loss": 1.04051986977458,
      "vision_logit_loss": 3.499508198797703
    },
    {
      "affinity_loss": 0.006614592779660597,
      "emkd_kd_loss": 1.5957689294591546,
      "epoch": 0.31047919736931817,
      "grad_norm": 0.3264337480068207,
      "hard_loss": 0.8357108208113277,
      "joint_kd_loss": 4.241403991468251,
      "learning_rate": 8.076760478866573e-06,
      "loss": 3.0887109375,
      "matched_vision_samples": 0.96375,
      "response_kd_loss": 1.0774518171545515,
      "sre_geom_loss": 0.002207131070582271,
      "sre_kd_loss": 6.8870391128538175,
      "sre_logit_loss": 6.709873561667628,
      "sre_span_loss": 0.0668090866995044,
      "step": 3700,
      "vision_kd_loss": 1.0570430210977793,
      "vision_logit_loss": 3.566712803840637
    },
    {
      "affinity_loss": 0.006449224810639862,
      "emkd_kd_loss": 1.548955143392086,
      "epoch": 0.3146748621986333,
      "grad_norm": 0.797814667224884,
      "hard_loss": 0.9127606771656065,
      "joint_kd_loss": 5.67656677082181,
      "learning_rate": 8.022923417744832e-06,
      "loss": 2.9068658447265623,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0403409866019957,
      "sre_geom_loss": 0.002263627827291721,
      "sre_kd_loss": 9.804178342660888,
      "sre_logit_loss": 9.622481892289361,
      "sre_span_loss": 0.06851501991972327,
      "step": 3750,
      "vision_kd_loss": 1.0287846502661706,
      "vision_logit_loss": 3.4702161172032358
    },
    {
      "affinity_loss": 0.006493313016253523,
      "emkd_kd_loss": 1.5770658253878356,
      "epoch": 0.3188705270279484,
      "grad_norm": 0.4198627769947052,
      "hard_loss": 0.9016739796711772,
      "joint_kd_loss": 4.865345234721899,
      "learning_rate": 7.968528214823232e-06,
      "loss": 2.7424136352539064,
      "matched_vision_samples": 0.94125,
      "response_kd_loss": 1.0995977397281285,
      "sre_geom_loss": 0.002321261092785534,
      "sre_kd_loss": 8.153624651189894,
      "sre_logit_loss": 7.970310364941834,
      "sre_span_loss": 0.06725126392440871,
      "step": 3800,
      "vision_kd_loss": 1.027266956269741,
      "vision_logit_loss": 3.459736520051956
    },
    {
      "affinity_loss": 0.0063233933618175794,
      "emkd_kd_loss": 1.5444312203675508,
      "epoch": 0.3230661918572635,
      "grad_norm": 0.4502379596233368,
      "hard_loss": 0.871540037859595,
      "joint_kd_loss": 4.751810603700578,
      "learning_rate": 7.913584913438056e-06,
      "loss": 2.9485772705078124,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.026945721456143,
      "sre_geom_loss": 0.0021011584640338297,
      "sre_kd_loss": 7.959189990526065,
      "sre_logit_loss": 7.7880371244577695,
      "sre_span_loss": 0.06609496979275718,
      "step": 3850,
      "vision_kd_loss": 1.0309583553671837,
      "vision_logit_loss": 3.491494080722332
    },
    {
      "affinity_loss": 0.006425961878558155,
      "emkd_kd_loss": 1.567815348431468,
      "epoch": 0.3272618566865786,
      "grad_norm": 0.1559724509716034,
      "hard_loss": 0.8896497608644768,
      "joint_kd_loss": 3.793287143856287,
      "learning_rate": 7.858103658124546e-06,
      "loss": 2.690289306640625,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.07362255421147,
      "sre_geom_loss": 0.002127243479800214,
      "sre_kd_loss": 6.018758946293965,
      "sre_logit_loss": 5.846681362219388,
      "sre_span_loss": 0.06571539437165484,
      "step": 3900,
      "vision_kd_loss": 1.0310040707141161,
      "vision_logit_loss": 3.4814200991392137
    },
    {
      "affinity_loss": 0.006445497672830243,
      "emkd_kd_loss": 1.5080766906030476,
      "epoch": 0.3314575215158937,
      "grad_norm": 0.5606459975242615,
      "hard_loss": 0.8852218116264158,
      "joint_kd_loss": 4.0063080033659935,
      "learning_rate": 7.802094692743825e-06,
      "loss": 2.780068359375,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 0.9247623250150013,
      "sre_geom_loss": 0.002043810671587778,
      "sre_kd_loss": 6.504539305600337,
      "sre_logit_loss": 6.336020742322144,
      "sre_span_loss": 0.06632803410990164,
      "step": 3950,
      "vision_kd_loss": 1.045695527419448,
      "vision_logit_loss": 3.5382323426008226
    },
    {
      "affinity_loss": 0.006370435900171287,
      "emkd_kd_loss": 1.4927999436110257,
      "epoch": 0.3356531863452088,
      "grad_norm": 0.2592237889766693,
      "hard_loss": 0.9088903229608332,
      "joint_kd_loss": 4.906821657307446,
      "learning_rate": 7.745568358591534e-06,
      "loss": 2.7466619873046874,
      "matched_vision_samples": 0.93125,
      "response_kd_loss": 0.9511421958217329,
      "sre_geom_loss": 0.0023100038299477886,
      "sre_kd_loss": 8.32084338221699,
      "sre_logit_loss": 8.136592017372314,
      "sre_span_loss": 0.06875117112882435,
      "step": 4000,
      "vision_kd_loss": 1.017228844985366,
      "vision_logit_loss": 3.4318717879056932
    },
    {
      "affinity_loss": 0.006574346447305288,
      "emkd_kd_loss": 1.5285608117096126,
      "epoch": 0.3398488511745239,
      "grad_norm": 0.3321019411087036,
      "hard_loss": 0.9300633328182357,
      "joint_kd_loss": 4.991243973933161,
      "learning_rate": 7.688535092488432e-06,
      "loss": 2.6139825439453124,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 0.9745684312691264,
      "sre_geom_loss": 0.002251646805565315,
      "sre_kd_loss": 8.453927140664309,
      "sre_logit_loss": 8.275197297860577,
      "sre_span_loss": 0.06614756843773648,
      "step": 4050,
      "vision_kd_loss": 1.0412765967100859,
      "vision_logit_loss": 3.507671732902527
    },
    {
      "affinity_loss": 0.006306191787007265,
      "emkd_kd_loss": 1.5231293231248855,
      "epoch": 0.344044516003839,
      "grad_norm": 0.3513050675392151,
      "hard_loss": 0.8723199477991693,
      "joint_kd_loss": 4.762742905467749,
      "learning_rate": 7.631005424853385e-06,
      "loss": 3.0026165771484377,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 0.9906912031493903,
      "sre_geom_loss": 0.0020528635434382634,
      "sre_kd_loss": 8.002356441854499,
      "sre_logit_loss": 7.834759908941342,
      "sre_span_loss": 0.06495341835310682,
      "step": 4100,
      "vision_kd_loss": 1.0277837195247412,
      "vision_logit_loss": 3.480515699982643
    },
    {
      "affinity_loss": 0.006252992601366714,
      "emkd_kd_loss": 1.530011322647333,
      "epoch": 0.34824018083315417,
      "grad_norm": 0.24557720124721527,
      "hard_loss": 0.8930681279379314,
      "joint_kd_loss": 5.946828401573002,
      "learning_rate": 7.572989977759073e-06,
      "loss": 2.8433306884765623,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0060459444111824,
      "sre_geom_loss": 0.0019248949580148178,
      "sre_kd_loss": 10.363645427315497,
      "sre_logit_loss": 10.201970050358504,
      "sre_span_loss": 0.06543058815645054,
      "step": 4150,
      "vision_kd_loss": 1.0269883524626493,
      "vision_logit_loss": 3.482654147744179
    },
    {
      "affinity_loss": 0.006484647126926575,
      "emkd_kd_loss": 1.5688138392567634,
      "epoch": 0.35243584566246927,
      "grad_norm": 0.26777133345603943,
      "hard_loss": 0.935561972084688,
      "joint_kd_loss": 5.733834979459643,
      "learning_rate": 7.514499462970757e-06,
      "loss": 2.8114093017578123,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0603049181187498,
      "sre_geom_loss": 0.0024101986736593515,
      "sre_kd_loss": 9.898856073822826,
      "sre_logit_loss": 9.709536467381986,
      "sre_span_loss": 0.06880973874824121,
      "step": 4200,
      "vision_kd_loss": 1.0386613804101943,
      "vision_logit_loss": 3.5061808103322982
    },
    {
      "affinity_loss": 0.006339285113790538,
      "emkd_kd_loss": 1.588299674540758,
      "epoch": 0.3566315104917844,
      "grad_norm": 0.2289292812347412,
      "hard_loss": 0.890563583004623,
      "joint_kd_loss": 5.163874879851937,
      "learning_rate": 7.455544679968507e-06,
      "loss": 2.710840148925781,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.1179736346087703,
      "sre_geom_loss": 0.001953855586469899,
      "sre_kd_loss": 8.739450063481927,
      "sre_logit_loss": 8.576566557739861,
      "sre_span_loss": 0.06519080239115282,
      "step": 4250,
      "vision_kd_loss": 1.0293128591030836,
      "vision_logit_loss": 3.4833229249715805
    },
    {
      "affinity_loss": 0.006540821462112945,
      "emkd_kd_loss": 1.5731028368696571,
      "epoch": 0.3608271753210995,
      "grad_norm": 0.48577117919921875,
      "hard_loss": 0.8977135098244616,
      "joint_kd_loss": 5.447506782226265,
      "learning_rate": 7.39613651395322e-06,
      "loss": 3.2296209716796875,
      "matched_vision_samples": 0.96,
      "response_kd_loss": 1.0425385059822065,
      "sre_geom_loss": 0.002173752893696701,
      "sre_kd_loss": 9.321910733697004,
      "sre_logit_loss": 9.147987079073792,
      "sre_span_loss": 0.06523602242348715,
      "step": 4300,
      "vision_kd_loss": 1.0518335826694964,
      "vision_logit_loss": 3.5532521864771844
    },
    {
      "affinity_loss": 0.006464149777311832,
      "emkd_kd_loss": 1.5224055562913419,
      "epoch": 0.3650228401504146,
      "grad_norm": 0.38548538088798523,
      "hard_loss": 0.8817649171372978,
      "joint_kd_loss": 3.8994488413259387,
      "learning_rate": 7.336285933836812e-06,
      "loss": 2.789610595703125,
      "matched_vision_samples": 0.93875,
      "response_kd_loss": 0.9727919269006361,
      "sre_geom_loss": 0.0022574435939196748,
      "sre_kd_loss": 6.276492160444613,
      "sre_logit_loss": 6.096237153696129,
      "sre_span_loss": 0.06738284743856639,
      "step": 4350,
      "vision_kd_loss": 1.0360095920413732,
      "vision_logit_loss": 3.497623391151428
    },
    {
      "affinity_loss": 0.006417335154837929,
      "emkd_kd_loss": 1.5879635883122682,
      "epoch": 0.3692185049797297,
      "grad_norm": 0.168227419257164,
      "hard_loss": 0.8942571031969783,
      "joint_kd_loss": 4.635253589972853,
      "learning_rate": 7.276003990216965e-06,
      "loss": 2.7563818359375,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0933877092978583,
      "sre_geom_loss": 0.0019509638495532045,
      "sre_kd_loss": 7.682543532769196,
      "sre_logit_loss": 7.520641348610516,
      "sre_span_loss": 0.06435396361863241,
      "step": 4400,
      "vision_kd_loss": 1.0412697314471007,
      "vision_logit_loss": 3.523345405757427
    },
    {
      "affinity_loss": 0.006509842612722423,
      "emkd_kd_loss": 1.5722032371908425,
      "epoch": 0.3734141698090448,
      "grad_norm": 0.38628196716308594,
      "hard_loss": 0.8891128304398443,
      "joint_kd_loss": 7.208361953310669,
      "learning_rate": 7.215301813336779e-06,
      "loss": 2.871839599609375,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0751621636929416,
      "sre_geom_loss": 0.0020700571286130297,
      "sre_kd_loss": 12.844520601043477,
      "sre_logit_loss": 12.674505984251155,
      "sre_span_loss": 0.0665116896899417,
      "step": 4450,
      "vision_kd_loss": 1.0346221570670604,
      "vision_logit_loss": 3.487504364848137
    },
    {
      "affinity_loss": 0.0064241454287548546,
      "emkd_kd_loss": 1.5454027200117708,
      "epoch": 0.3776098346383599,
      "grad_norm": 0.3604796230792999,
      "hard_loss": 0.8472890145601923,
      "joint_kd_loss": 4.372796813733876,
      "learning_rate": 7.154190611029724e-06,
      "loss": 2.750163879394531,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0102250605963263,
      "sre_geom_loss": 0.0021002317212867184,
      "sre_kd_loss": 7.200190916303545,
      "sre_logit_loss": 7.0304852031706835,
      "sre_span_loss": 0.06469413840211928,
      "step": 4500,
      "vision_kd_loss": 1.0402901881933213,
      "vision_logit_loss": 3.518746203184128
    },
    {
      "affinity_loss": 0.00626316822192166,
      "emkd_kd_loss": 1.524771582596004,
      "epoch": 0.381805499467675,
      "grad_norm": 0.2859608829021454,
      "hard_loss": 0.8843401169731078,
      "joint_kd_loss": 6.047029048129916,
      "learning_rate": 7.092681666650267e-06,
      "loss": 3.109582824707031,
      "matched_vision_samples": 0.9375,
      "response_kd_loss": 1.0040588567053192,
      "sre_geom_loss": 0.0021020607675961856,
      "sre_kd_loss": 10.569286539112218,
      "sre_logit_loss": 10.397673731856049,
      "sre_span_loss": 0.06650968992849812,
      "step": 4550,
      "vision_kd_loss": 1.022742153853178,
      "vision_logit_loss": 3.4646517941355706
    },
    {
      "affinity_loss": 0.006568131280364469,
      "emkd_kd_loss": 1.538759891502559,
      "epoch": 0.38600116429699016,
      "grad_norm": 0.24783626198768616,
      "hard_loss": 0.9132439545560191,
      "joint_kd_loss": 4.331208887957036,
      "learning_rate": 7.03078633699055e-06,
      "loss": 2.483944549560547,
      "matched_vision_samples": 0.95875,
      "response_kd_loss": 0.9729009878324724,
      "sre_geom_loss": 0.0018245813964151126,
      "sre_kd_loss": 7.123657841975801,
      "sre_logit_loss": 6.966566690816544,
      "sre_span_loss": 0.06586203658487648,
      "step": 4600,
      "vision_kd_loss": 1.0523093967884778,
      "vision_logit_loss": 3.552424457371235
    },
    {
      "affinity_loss": 0.006412309370061848,
      "emkd_kd_loss": 1.5179644768685103,
      "epoch": 0.39019682912630527,
      "grad_norm": 0.22662273049354553,
      "hard_loss": 0.8502585530698571,
      "joint_kd_loss": 5.157194656655192,
      "learning_rate": 6.968516050183522e-06,
      "loss": 3.159313049316406,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 0.9629473331730151,
      "sre_geom_loss": 0.002348349618746397,
      "sre_kd_loss": 8.796424812832846,
      "sre_logit_loss": 8.612505841175734,
      "sre_span_loss": 0.06650151721201837,
      "step": 4650,
      "vision_kd_loss": 1.0364908103644848,
      "vision_logit_loss": 3.5047323006391524
    },
    {
      "affinity_loss": 0.006277053122757934,
      "emkd_kd_loss": 1.5483670952171087,
      "epoch": 0.39439249395562037,
      "grad_norm": 0.5042697191238403,
      "hard_loss": 0.8892669618060427,
      "joint_kd_loss": 5.19815808199346,
      "learning_rate": 6.905882303592875e-06,
      "loss": 2.9537255859375,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0317133860014291,
      "sre_geom_loss": 0.002243184985501898,
      "sre_kd_loss": 8.847949106688612,
      "sre_logit_loss": 8.670905912161107,
      "sre_span_loss": 0.06488399909809232,
      "step": 4700,
      "vision_kd_loss": 1.0325104007869959,
      "vision_logit_loss": 3.5023362877964974
    },
    {
      "affinity_loss": 0.006585628016036935,
      "emkd_kd_loss": 1.5564341542124749,
      "epoch": 0.3985881587849355,
      "grad_norm": 0.4606955349445343,
      "hard_loss": 0.8622168833888282,
      "joint_kd_loss": 3.894696131721139,
      "learning_rate": 6.8428966616902215e-06,
      "loss": 2.7049560546875,
      "matched_vision_samples": 0.9625,
      "response_kd_loss": 1.0089894844066953,
      "sre_geom_loss": 0.0018542648446882026,
      "sre_kd_loss": 6.232958115222864,
      "sre_logit_loss": 6.076307926937879,
      "sre_span_loss": 0.06393692264566198,
      "step": 4750,
      "vision_kd_loss": 1.0519394133985043,
      "vision_logit_loss": 3.5491948571801184
    },
    {
      "affinity_loss": 0.006316313436545897,
      "emkd_kd_loss": 1.5606539480760693,
      "epoch": 0.4027838236142506,
      "grad_norm": 0.5023351907730103,
      "hard_loss": 0.902390360827485,
      "joint_kd_loss": 3.463325775153935,
      "learning_rate": 6.779570753919874e-06,
      "loss": 2.3669093322753905,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.0383148521683687,
      "sre_geom_loss": 0.0019649476642169183,
      "sre_kd_loss": 5.365997604532168,
      "sre_logit_loss": 5.20472462102247,
      "sre_span_loss": 0.06302559396717697,
      "step": 4800,
      "vision_kd_loss": 1.0414965204149484,
      "vision_logit_loss": 3.534354736506939
    },
    {
      "affinity_loss": 0.006259767573210411,
      "emkd_kd_loss": 1.5389564171060919,
      "epoch": 0.4069794884435657,
      "grad_norm": 0.4994589388370514,
      "hard_loss": 0.8612965436837112,
      "joint_kd_loss": 4.639349498637021,
      "learning_rate": 6.715916272551616e-06,
      "loss": 2.8855938720703125,
      "matched_vision_samples": 0.94125,
      "response_kd_loss": 1.030290936301044,
      "sre_geom_loss": 0.0020937916578972703,
      "sre_kd_loss": 7.739742590212263,
      "sre_logit_loss": 7.569307433088543,
      "sre_span_loss": 0.06574557672953234,
      "step": 4850,
      "vision_kd_loss": 1.0238109508901834,
      "vision_logit_loss": 3.4692670515179636
    },
    {
      "affinity_loss": 0.00656244074110873,
      "emkd_kd_loss": 1.5627345988340675,
      "epoch": 0.4111751532728808,
      "grad_norm": 0.4659096896648407,
      "hard_loss": 0.8974591805158343,
      "joint_kd_loss": 4.48880862571299,
      "learning_rate": 6.65194497052189e-06,
      "loss": 2.6385205078125,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0328272329373795,
      "sre_geom_loss": 0.0018637150466705066,
      "sre_kd_loss": 7.414882560810074,
      "sre_logit_loss": 7.256829242795939,
      "sre_span_loss": 0.06486755991354584,
      "step": 4900,
      "vision_kd_loss": 1.0463209812343122,
      "vision_logit_loss": 3.5290398606657982
    },
    {
      "affinity_loss": 0.006447013877332211,
      "emkd_kd_loss": 1.5758402932807802,
      "epoch": 0.4153708181021959,
      "grad_norm": 0.24411611258983612,
      "hard_loss": 0.8622441565912595,
      "joint_kd_loss": 4.358850977569818,
      "learning_rate": 6.587668659263775e-06,
      "loss": 2.9522540283203127,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.064698241759088,
      "sre_geom_loss": 0.0020893224239121365,
      "sre_kd_loss": 7.141861667283811,
      "sre_logit_loss": 6.9713621329009765,
      "sre_span_loss": 0.06603347484953702,
      "step": 4950,
      "vision_kd_loss": 1.0434911727160214,
      "vision_logit_loss": 3.5292632964253428
    },
    {
      "affinity_loss": 0.006402284058567602,
      "emkd_kd_loss": 1.5764398024231197,
      "epoch": 0.419566482931511,
      "grad_norm": 0.2983902096748352,
      "hard_loss": 0.8885921435094406,
      "joint_kd_loss": 5.19882812127471,
      "learning_rate": 6.5230992065261685e-06,
      "loss": 2.654967041015625,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 1.0677396131609862,
      "sre_geom_loss": 0.002139802429877875,
      "sre_kd_loss": 8.821216418966651,
      "sre_logit_loss": 8.64900579128618,
      "sre_span_loss": 0.06522052363492549,
      "step": 5000,
      "vision_kd_loss": 1.04256999284029,
      "vision_logit_loss": 3.5300515696406363
    },
    {
      "affinity_loss": 0.006325366868550191,
      "emkd_kd_loss": 1.526077223457396,
      "epoch": 0.4237621477608261,
      "grad_norm": 0.13848425447940826,
      "hard_loss": 0.8743577154123432,
      "joint_kd_loss": 5.593208262901753,
      "learning_rate": 6.458248534182565e-06,
      "loss": 2.808915710449219,
      "matched_vision_samples": 0.945,
      "response_kd_loss": 0.9852503161493302,
      "sre_geom_loss": 0.002286595537597318,
      "sre_kd_loss": 9.660339315840975,
      "sre_logit_loss": 9.48047980551928,
      "sre_span_loss": 0.06552975815720856,
      "step": 5050,
      "vision_kd_loss": 1.0334520656242967,
      "vision_logit_loss": 3.501271575987339
    },
    {
      "affinity_loss": 0.006433573642279953,
      "emkd_kd_loss": 1.5413343020668253,
      "epoch": 0.42795781259014126,
      "grad_norm": 0.09087979048490524,
      "hard_loss": 0.86252485414484,
      "joint_kd_loss": 4.401649188641459,
      "learning_rate": 6.393128616029845e-06,
      "loss": 2.9805877685546873,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0164900969852908,
      "sre_geom_loss": 0.001949118218560102,
      "sre_kd_loss": 7.261964121309575,
      "sre_logit_loss": 7.099827310492255,
      "sre_span_loss": 0.06468086811364629,
      "step": 5100,
      "vision_kd_loss": 1.0330892534926535,
      "vision_logit_loss": 3.4889996539056303
    },
    {
      "affinity_loss": 0.006436176425631856,
      "emkd_kd_loss": 1.55717256359756,
      "epoch": 0.43215347741945637,
      "grad_norm": 0.5038453936576843,
      "hard_loss": 0.8796391276036957,
      "joint_kd_loss": 4.108262536972761,
      "learning_rate": 6.3277514755774665e-06,
      "loss": 2.7804132080078126,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.039129797567204,
      "sre_geom_loss": 0.001972435232401253,
      "sre_kd_loss": 6.659352532711345,
      "sre_logit_loss": 6.496392800107497,
      "sre_span_loss": 0.06433793148607947,
      "step": 5150,
      "vision_kd_loss": 1.037607665285468,
      "vision_logit_loss": 3.5068130199611187
    },
    {
      "affinity_loss": 0.006498614853335312,
      "emkd_kd_loss": 1.5534534925129264,
      "epoch": 0.43634914224877147,
      "grad_norm": 0.3289719820022583,
      "hard_loss": 0.8612535946663956,
      "joint_kd_loss": 3.7395319766178727,
      "learning_rate": 6.262129183827507e-06,
      "loss": 2.8938897705078124,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0047027064979046,
      "sre_geom_loss": 0.0020866511379728616,
      "sre_kd_loss": 5.925610470755491,
      "sre_logit_loss": 5.757337604991335,
      "sre_span_loss": 0.06394026616821065,
      "step": 5200,
      "vision_kd_loss": 1.051102136299014,
      "vision_logit_loss": 3.5545470620691777
    },
    {
      "affinity_loss": 0.006505489894625498,
      "emkd_kd_loss": 1.5499990355037152,
      "epoch": 0.4405448070780866,
      "grad_norm": 0.26362961530685425,
      "hard_loss": 0.8925863837557881,
      "joint_kd_loss": 4.779184609632939,
      "learning_rate": 6.196273857045901e-06,
      "loss": 2.7301348876953124,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.0079814199499213,
      "sre_geom_loss": 0.0020805374352625174,
      "sre_kd_loss": 8.008370190088172,
      "sre_logit_loss": 7.838656875397719,
      "sre_span_loss": 0.06568647527834401,
      "step": 5250,
      "vision_kd_loss": 1.0460083260014654,
      "vision_logit_loss": 3.5334843184053897
    },
    {
      "affinity_loss": 0.006418626114609651,
      "emkd_kd_loss": 1.5477352772559971,
      "epoch": 0.4447404719074017,
      "grad_norm": 0.7957174181938171,
      "hard_loss": 0.8666706612517032,
      "joint_kd_loss": 4.277622215002776,
      "learning_rate": 6.13019765452534e-06,
      "loss": 2.5488577270507813,
      "matched_vision_samples": 0.956875,
      "response_kd_loss": 1.000248505110544,
      "sre_geom_loss": 0.002048450085843143,
      "sre_kd_loss": 7.007509178705513,
      "sre_logit_loss": 6.842223455331914,
      "sre_span_loss": 0.06286319745471701,
      "step": 5300,
      "vision_kd_loss": 1.0476110274344683,
      "vision_logit_loss": 3.548581501543522
    },
    {
      "affinity_loss": 0.0063232823212456424,
      "emkd_kd_loss": 1.5335007371380924,
      "epoch": 0.4489361367367168,
      "grad_norm": 0.15361563861370087,
      "hard_loss": 0.9179482305182227,
      "joint_kd_loss": 4.276777437608689,
      "learning_rate": 6.063912776340224e-06,
      "loss": 2.4594891357421873,
      "matched_vision_samples": 0.940625,
      "response_kd_loss": 1.018081890339563,
      "sre_geom_loss": 0.0020219750780461256,
      "sre_kd_loss": 7.020054106251337,
      "sre_logit_loss": 6.852465295452566,
      "sre_span_loss": 0.0664900626335293,
      "step": 5350,
      "vision_kd_loss": 1.0244597905874253,
      "vision_logit_loss": 3.465510929375887
    },
    {
      "affinity_loss": 0.006452494801487773,
      "emkd_kd_loss": 1.5260340528283267,
      "epoch": 0.4531318015660319,
      "grad_norm": 0.18361859023571014,
      "hard_loss": 0.8920993885313828,
      "joint_kd_loss": 3.8770836304873226,
      "learning_rate": 5.997431461094084e-06,
      "loss": 2.4538299560546877,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 0.9585883375892195,
      "sre_geom_loss": 0.0019811052383585446,
      "sre_kd_loss": 6.228133177671116,
      "sre_logit_loss": 6.065482758810976,
      "sre_span_loss": 0.06359517352888361,
      "step": 5400,
      "vision_kd_loss": 1.046739885881543,
      "vision_logit_loss": 3.5417100594937803
    },
    {
      "affinity_loss": 0.0065755589051696,
      "emkd_kd_loss": 1.5679663044586778,
      "epoch": 0.457327466395347,
      "grad_norm": 0.13658566772937775,
      "hard_loss": 0.9083725026540742,
      "joint_kd_loss": 4.9667519610747695,
      "learning_rate": 5.930765983659888e-06,
      "loss": 2.978166198730469,
      "matched_vision_samples": 0.955625,
      "response_kd_loss": 1.0345028904862204,
      "sre_geom_loss": 0.0022244674197148063,
      "sre_kd_loss": 8.365537614952773,
      "sre_logit_loss": 8.189552739478676,
      "sre_span_loss": 0.0647615263343323,
      "step": 5450,
      "vision_kd_loss": 1.050714857839048,
      "vision_logit_loss": 3.5453035463392735
    },
    {
      "affinity_loss": 0.006439475532970391,
      "emkd_kd_loss": 1.5282507465034723,
      "epoch": 0.4615231312246621,
      "grad_norm": 0.0942656397819519,
      "hard_loss": 0.9191378186881775,
      "joint_kd_loss": 4.796469546090812,
      "learning_rate": 5.8639286529136555e-06,
      "loss": 2.4882350158691406,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 1.0003190245024292,
      "sre_geom_loss": 0.0021448079797037867,
      "sre_kd_loss": 8.064688357339474,
      "sre_logit_loss": 7.892415539580834,
      "sre_span_loss": 0.06503244865918532,
      "step": 5500,
      "vision_kd_loss": 1.0280912338197232,
      "vision_logit_loss": 3.468417379260063
    },
    {
      "affinity_loss": 0.006530446990509518,
      "emkd_kd_loss": 1.5341182498447596,
      "epoch": 0.4657187960539772,
      "grad_norm": 0.136615589261055,
      "hard_loss": 0.8208159855576923,
      "joint_kd_loss": 3.561297087799758,
      "learning_rate": 5.796931809461788e-06,
      "loss": 2.6132183837890626,
      "matched_vision_samples": 0.963125,
      "response_kd_loss": 0.9586799608741717,
      "sre_geom_loss": 0.0019663021957154834,
      "sre_kd_loss": 5.588475915235467,
      "sre_logit_loss": 5.426860242538969,
      "sre_span_loss": 0.06330054241931066,
      "step": 5550,
      "vision_kd_loss": 1.054778268262744,
      "vision_logit_loss": 3.5660683761537073
    },
    {
      "affinity_loss": 0.006360625254164916,
      "emkd_kd_loss": 1.5305048801284284,
      "epoch": 0.46991446088329236,
      "grad_norm": 0.35967713594436646,
      "hard_loss": 0.9158249749731294,
      "joint_kd_loss": 5.563630853388458,
      "learning_rate": 5.729787823362547e-06,
      "loss": 2.8312060546875,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.0087424529840485,
      "sre_geom_loss": 0.0021686411360275315,
      "sre_kd_loss": 9.596756726335734,
      "sre_logit_loss": 9.424075531303679,
      "sre_span_loss": 0.06424917083000764,
      "step": 5600,
      "vision_kd_loss": 1.0261336542293429,
      "vision_logit_loss": 3.4684720860421656
    },
    {
      "affinity_loss": 0.006400668884743936,
      "emkd_kd_loss": 1.55388119609328,
      "epoch": 0.47411012571260747,
      "grad_norm": 0.20425842702388763,
      "hard_loss": 0.8970619056722399,
      "joint_kd_loss": 4.267423912901432,
      "learning_rate": 5.662509091842081e-06,
      "loss": 2.5707601928710937,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.0499878677068921,
      "sre_geom_loss": 0.002112204063688097,
      "sre_kd_loss": 6.980966597697698,
      "sre_logit_loss": 6.81106738805669,
      "sre_span_loss": 0.06428897679783403,
      "step": 5650,
      "vision_kd_loss": 1.0288872618600726,
      "vision_logit_loss": 3.475482160151005
    },
    {
      "affinity_loss": 0.006373689894971903,
      "emkd_kd_loss": 1.5234981606714428,
      "epoch": 0.47830579054192257,
      "grad_norm": 0.08710998296737671,
      "hard_loss": 0.8762370741959512,
      "joint_kd_loss": 4.564999839495868,
      "learning_rate": 5.5951080370054585e-06,
      "loss": 2.7600198364257813,
      "matched_vision_samples": 0.955625,
      "response_kd_loss": 0.9670912751605641,
      "sre_geom_loss": 0.0019500405732396243,
      "sre_kd_loss": 7.606501507470385,
      "sre_logit_loss": 7.4464900953054896,
      "sre_span_loss": 0.06250936713069677,
      "step": 5700,
      "vision_kd_loss": 1.0399525224417447,
      "vision_logit_loss": 3.5224410992860795
    },
    {
      "affinity_loss": 0.00636388226528652,
      "emkd_kd_loss": 1.568584161219187,
      "epoch": 0.4825014553712377,
      "grad_norm": 0.4114396870136261,
      "hard_loss": 0.9189355904477998,
      "joint_kd_loss": 5.523851202446967,
      "learning_rate": 5.527597103543087e-06,
      "loss": 2.8405752563476563,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 1.083838328427634,
      "sre_geom_loss": 0.0020338168593829666,
      "sre_kd_loss": 9.479118208414874,
      "sre_logit_loss": 9.311749940580775,
      "sre_span_loss": 0.06567740591941401,
      "step": 5750,
      "vision_kd_loss": 1.026664997525513,
      "vision_logit_loss": 3.4702717697620393
    },
    {
      "affinity_loss": 0.006448363448434975,
      "emkd_kd_loss": 1.5320797955500893,
      "epoch": 0.4866971202005528,
      "grad_norm": 0.21473295986652374,
      "hard_loss": 0.8276228061778693,
      "joint_kd_loss": 4.813683888502419,
      "learning_rate": 5.459988756432972e-06,
      "loss": 2.6255874633789062,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 0.9738771197747542,
      "sre_geom_loss": 0.0018583660461598938,
      "sre_kd_loss": 8.095287995019461,
      "sre_logit_loss": 7.9392166676805935,
      "sre_span_loss": 0.0631530747609213,
      "step": 5800,
      "vision_kd_loss": 1.0451412354409695,
      "vision_logit_loss": 3.5357285994291305
    },
    {
      "affinity_loss": 0.0064296321119763885,
      "emkd_kd_loss": 1.511286385562271,
      "epoch": 0.4908927850298679,
      "grad_norm": 0.14952680468559265,
      "hard_loss": 0.8822379581958648,
      "joint_kd_loss": 3.4354052251018583,
      "learning_rate": 5.392295478639226e-06,
      "loss": 2.4776400756835937,
      "matched_vision_samples": 0.9625,
      "response_kd_loss": 0.9435495138556758,
      "sre_geom_loss": 0.0018984888688479628,
      "sre_kd_loss": 5.359524062720593,
      "sre_logit_loss": 5.201478142952765,
      "sre_span_loss": 0.06312151078600436,
      "step": 5850,
      "vision_kd_loss": 1.039511627405882,
      "vision_logit_loss": 3.515083299577236
    },
    {
      "affinity_loss": 0.006542119915829972,
      "emkd_kd_loss": 1.5409159213956445,
      "epoch": 0.495088449859183,
      "grad_norm": 0.2927500009536743,
      "hard_loss": 0.8736963405211146,
      "joint_kd_loss": 4.053891328610479,
      "learning_rate": 5.3245297688072605e-06,
      "loss": 2.6481338500976563,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.0040678097290174,
      "sre_geom_loss": 0.0020597887723457563,
      "sre_kd_loss": 6.566866683678236,
      "sre_logit_loss": 6.399609629756596,
      "sre_span_loss": 0.06426762644201517,
      "step": 5900,
      "vision_kd_loss": 1.038882015645504,
      "vision_logit_loss": 3.5013160744309424
    },
    {
      "affinity_loss": 0.006412227299733786,
      "emkd_kd_loss": 1.5577597698941827,
      "epoch": 0.4992841146884981,
      "grad_norm": 0.19900836050510406,
      "hard_loss": 0.8377585814029817,
      "joint_kd_loss": 4.137937448453158,
      "learning_rate": 5.256704138956082e-06,
      "loss": 2.4006602478027346,
      "matched_vision_samples": 0.958125,
      "response_kd_loss": 1.036087076849293,
      "sre_geom_loss": 0.0020628783980055233,
      "sre_kd_loss": 6.718115149359218,
      "sre_logit_loss": 6.550383211429725,
      "sre_span_loss": 0.06458802795154042,
      "step": 5950,
      "vision_kd_loss": 1.039716230146587,
      "vision_logit_loss": 3.5176421885192393
    },
    {
      "affinity_loss": 0.006562388676538831,
      "emkd_kd_loss": 1.6011878018453718,
      "epoch": 0.5034797795178132,
      "grad_norm": 0.08337504416704178,
      "hard_loss": 0.8904860830825418,
      "joint_kd_loss": 4.912277173958719,
      "learning_rate": 5.188831112168115e-06,
      "loss": 2.709776611328125,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.1146977969957517,
      "sre_geom_loss": 0.002019169788127755,
      "sre_kd_loss": 8.223366517652757,
      "sre_logit_loss": 8.057020596333896,
      "sre_span_loss": 0.06538749206345529,
      "step": 6000,
      "vision_kd_loss": 1.0438389027118682,
      "vision_logit_loss": 3.519116739332676
    },
    {
      "affinity_loss": 0.006453795163106406,
      "emkd_kd_loss": 1.5489587873686106,
      "epoch": 0.5076754443471283,
      "grad_norm": 0.18213827908039093,
      "hard_loss": 0.9155912608149811,
      "joint_kd_loss": 4.4431769328005615,
      "learning_rate": 5.120923220276984e-06,
      "loss": 2.6758682250976564,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.0214847806454384,
      "sre_geom_loss": 0.0020776605379163727,
      "sre_kd_loss": 7.337395086940378,
      "sre_logit_loss": 7.1694014127500125,
      "sre_span_loss": 0.0641106352233328,
      "step": 6050,
      "vision_kd_loss": 1.0382163950428367,
      "vision_logit_loss": 3.5074860627949236
    },
    {
      "affinity_loss": 0.006485700674238615,
      "emkd_kd_loss": 1.5310127290477977,
      "epoch": 0.5118711091764434,
      "grad_norm": 0.13631512224674225,
      "hard_loss": 0.8950701212025797,
      "joint_kd_loss": 3.639416866125539,
      "learning_rate": 5.052993001553684e-06,
      "loss": 2.6089068603515626,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 0.9796935845838732,
      "sre_geom_loss": 0.0019328760783015931,
      "sre_kd_loss": 5.7478210172173565,
      "sre_logit_loss": 5.58812647155035,
      "sre_span_loss": 0.06305077654425986,
      "step": 6100,
      "vision_kd_loss": 1.0411659357324243,
      "vision_logit_loss": 3.516093674302101
    },
    {
      "affinity_loss": 0.0064450326596852394,
      "emkd_kd_loss": 1.573933630120009,
      "epoch": 0.5160667740057585,
      "grad_norm": 0.09373082220554352,
      "hard_loss": 0.8919830626021666,
      "joint_kd_loss": 4.870283726993948,
      "learning_rate": 4.985052998391553e-06,
      "loss": 2.5222955322265626,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0730119306170076,
      "sre_geom_loss": 0.0019892766410648656,
      "sre_kd_loss": 8.166633789930493,
      "sre_logit_loss": 8.00218194876259,
      "sre_span_loss": 0.06498804704868234,
      "step": 6150,
      "vision_kd_loss": 1.0374276673048735,
      "vision_logit_loss": 3.5052074028551576
    },
    {
      "affinity_loss": 0.006457659800071269,
      "emkd_kd_loss": 1.5365745543316007,
      "epoch": 0.5202624388350736,
      "grad_norm": 0.42389824986457825,
      "hard_loss": 0.8782488316192893,
      "joint_kd_loss": 5.55733553275466,
      "learning_rate": 4.917115754990497e-06,
      "loss": 3.0090460205078124,
      "matched_vision_samples": 0.955625,
      "response_kd_loss": 0.9927252908157379,
      "sre_geom_loss": 0.0020690439500447157,
      "sre_kd_loss": 9.578096460234374,
      "sre_logit_loss": 9.410116823319258,
      "sre_span_loss": 0.06452739921398461,
      "step": 6200,
      "vision_kd_loss": 1.040211909338832,
      "vision_logit_loss": 3.5150816568732264
    },
    {
      "affinity_loss": 0.006301528327458073,
      "emkd_kd_loss": 1.5265972565952688,
      "epoch": 0.5244581036643887,
      "grad_norm": 0.16932791471481323,
      "hard_loss": 0.855999487630852,
      "joint_kd_loss": 3.9054154692962766,
      "learning_rate": 4.849193815040869e-06,
      "loss": 2.4828466796875,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 0.9968601695634959,
      "sre_geom_loss": 0.0020865747483190945,
      "sre_kd_loss": 6.284233689984539,
      "sre_logit_loss": 6.116449048121576,
      "sre_span_loss": 0.06345588561496697,
      "step": 6250,
      "vision_kd_loss": 1.0281671724468469,
      "vision_logit_loss": 3.4825158561766147
    },
    {
      "affinity_loss": 0.006460933751659468,
      "emkd_kd_loss": 1.5726350755803287,
      "epoch": 0.5286537684937039,
      "grad_norm": 0.24956093728542328,
      "hard_loss": 0.8507355281633363,
      "joint_kd_loss": 4.830271871685982,
      "learning_rate": 4.781299719407445e-06,
      "loss": 2.5963726806640626,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 1.0599253409834375,
      "sre_geom_loss": 0.00200917945800569,
      "sre_kd_loss": 8.087908684976865,
      "sre_logit_loss": 7.9235527788387845,
      "sre_span_loss": 0.06389693847741,
      "step": 6300,
      "vision_kd_loss": 1.0426724069938063,
      "vision_logit_loss": 3.5245962524414063
    },
    {
      "affinity_loss": 0.006430318771890597,
      "emkd_kd_loss": 1.5515664682071657,
      "epoch": 0.532849433323019,
      "grad_norm": 0.11670966446399689,
      "hard_loss": 0.9150859561292509,
      "joint_kd_loss": 4.345141123663634,
      "learning_rate": 4.713446003813924e-06,
      "loss": 2.445212097167969,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 1.0406461312571706,
      "sre_geom_loss": 0.0020599929296488727,
      "sre_kd_loss": 7.1387157994741575,
      "sre_logit_loss": 6.97197154559166,
      "sre_span_loss": 0.06374461955856532,
      "step": 6350,
      "vision_kd_loss": 1.03124340493232,
      "vision_logit_loss": 3.481941743940115
    },
    {
      "affinity_loss": 0.006424195358849829,
      "emkd_kd_loss": 1.5371554645523429,
      "epoch": 0.5370450981523341,
      "grad_norm": 0.11709947139024734,
      "hard_loss": 0.9154249899749266,
      "joint_kd_loss": 4.249602195937187,
      "learning_rate": 4.645645196528386e-06,
      "loss": 2.587103271484375,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0072717313562998,
      "sre_geom_loss": 0.0019358596840132413,
      "sre_kd_loss": 6.962048908402212,
      "sre_logit_loss": 6.801298973062512,
      "sre_span_loss": 0.06395694629987701,
      "step": 6400,
      "vision_kd_loss": 1.033519599735737,
      "vision_logit_loss": 3.491658863276243
    },
    {
      "affinity_loss": 0.006394495793065289,
      "emkd_kd_loss": 1.5574661997426302,
      "epoch": 0.5412407629816492,
      "grad_norm": 0.377288281917572,
      "hard_loss": 0.9046337586716345,
      "joint_kd_loss": 4.322641098797321,
      "learning_rate": 4.5779098160501036e-06,
      "loss": 2.570833435058594,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.044978889366671,
      "sre_geom_loss": 0.001958895961975333,
      "sre_kd_loss": 7.08781594755128,
      "sre_logit_loss": 6.9264459213911325,
      "sre_span_loss": 0.06342522404505871,
      "step": 6450,
      "vision_kd_loss": 1.0349767533689738,
      "vision_logit_loss": 3.5004574358463287
    },
    {
      "affinity_loss": 0.006560055359441322,
      "emkd_kd_loss": 1.5523770747333765,
      "epoch": 0.5454364278109644,
      "grad_norm": 0.169332817196846,
      "hard_loss": 0.8482358703043519,
      "joint_kd_loss": 4.769595405701548,
      "learning_rate": 4.510252368798176e-06,
      "loss": 2.5276068115234374,
      "matched_vision_samples": 0.96125,
      "response_kd_loss": 1.0024479312822228,
      "sre_geom_loss": 0.0019922620212864217,
      "sre_kd_loss": 7.986813722562511,
      "sre_logit_loss": 7.822725031866575,
      "sre_span_loss": 0.06447566003189423,
      "step": 6500,
      "vision_kd_loss": 1.0511531107127667,
      "vision_logit_loss": 3.5486069050431253
    },
    {
      "affinity_loss": 0.006485574078833451,
      "emkd_kd_loss": 1.5382341915927826,
      "epoch": 0.5496320926402795,
      "grad_norm": 0.2081226259469986,
      "hard_loss": 0.8810288547346591,
      "joint_kd_loss": 5.4051060951687395,
      "learning_rate": 4.442685346802395e-06,
      "loss": 2.5120989990234377,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 0.9918361154746594,
      "sre_geom_loss": 0.0019487059076678291,
      "sre_kd_loss": 9.27197805538075,
      "sre_logit_loss": 9.11071920001952,
      "sre_span_loss": 0.06382356106652878,
      "step": 6550,
      "vision_kd_loss": 1.042316133007407,
      "vision_logit_loss": 3.520707120895386
    },
    {
      "affinity_loss": 0.0065279968713002744,
      "emkd_kd_loss": 1.5447081546485424,
      "epoch": 0.5538277574695946,
      "grad_norm": 0.23501549661159515,
      "hard_loss": 0.8490780742594325,
      "joint_kd_loss": 4.395601430349052,
      "learning_rate": 4.3752212253967505e-06,
      "loss": 2.5457908630371096,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.0080117493962137,
      "sre_geom_loss": 0.0019526363992783046,
      "sre_kd_loss": 7.246494700317271,
      "sre_logit_loss": 7.0853776412647855,
      "sre_span_loss": 0.06348523222142831,
      "step": 6600,
      "vision_kd_loss": 1.040702280215919,
      "vision_logit_loss": 3.5100094315409662
    },
    {
      "affinity_loss": 0.006540068430331303,
      "emkd_kd_loss": 1.5873305475525559,
      "epoch": 0.5580234222989097,
      "grad_norm": 0.11410835385322571,
      "hard_loss": 0.8595376335874789,
      "joint_kd_loss": 3.688180846106261,
      "learning_rate": 4.3078724609160385e-06,
      "loss": 2.516480255126953,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0824897612107571,
      "sre_geom_loss": 0.002039088936091158,
      "sre_kd_loss": 5.789031173668336,
      "sre_logit_loss": 5.622282287124835,
      "sre_span_loss": 0.064794447098393,
      "step": 6650,
      "vision_kd_loss": 1.0460856657102704,
      "vision_logit_loss": 3.530335822403431
    },
    {
      "affinity_loss": 0.006519050355855142,
      "emkd_kd_loss": 1.5401995258964598,
      "epoch": 0.5622190871282248,
      "grad_norm": 0.21049073338508606,
      "hard_loss": 0.8463549892129959,
      "joint_kd_loss": 5.419794214107096,
      "learning_rate": 4.240651488395958e-06,
      "loss": 2.753148498535156,
      "matched_vision_samples": 0.953125,
      "response_kd_loss": 0.9847610251529397,
      "sre_geom_loss": 0.00204836312389812,
      "sre_kd_loss": 9.299388912988361,
      "sre_logit_loss": 9.131990662705794,
      "sre_span_loss": 0.06498014155193232,
      "step": 6700,
      "vision_kd_loss": 1.0478190122544766,
      "vision_logit_loss": 3.539371015429497
    },
    {
      "affinity_loss": 0.006475101869436912,
      "emkd_kd_loss": 1.5620221250876785,
      "epoch": 0.5664147519575399,
      "grad_norm": 0.10555559396743774,
      "hard_loss": 0.858259772170095,
      "joint_kd_loss": 3.7625388849712906,
      "learning_rate": 4.173570719277165e-06,
      "loss": 2.5330686950683594,
      "matched_vision_samples": 0.958125,
      "response_kd_loss": 1.0448017074440332,
      "sre_geom_loss": 0.0021366895074292815,
      "sre_kd_loss": 5.963055643907283,
      "sre_logit_loss": 5.793303932358249,
      "sre_span_loss": 0.06291726480587385,
      "step": 6750,
      "vision_kd_loss": 1.039621271826327,
      "vision_logit_loss": 3.5109748999774455
    },
    {
      "affinity_loss": 0.006452931987587363,
      "emkd_kd_loss": 1.5327877719514071,
      "epoch": 0.570610416786855,
      "grad_norm": 0.5356866717338562,
      "hard_loss": 0.8674594598565818,
      "joint_kd_loss": 4.465606055874377,
      "learning_rate": 4.106642539113649e-06,
      "loss": 2.8544131469726564,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 0.9976677345474831,
      "sre_geom_loss": 0.0019471254867752031,
      "sre_kd_loss": 7.398424368938431,
      "sre_logit_loss": 7.236749020485004,
      "sre_span_loss": 0.06431906476966105,
      "step": 6800,
      "vision_kd_loss": 1.0339539047703148,
      "vision_logit_loss": 3.4905224262177943
    },
    {
      "affinity_loss": 0.0065213222468446476,
      "emkd_kd_loss": 1.5594844456994905,
      "epoch": 0.5748060816161701,
      "grad_norm": 0.2253926694393158,
      "hard_loss": 0.8953432530842838,
      "joint_kd_loss": 5.124367073588073,
      "learning_rate": 4.039879305285927e-06,
      "loss": 2.836796875,
      "matched_vision_samples": 0.951875,
      "response_kd_loss": 1.0344120261302725,
      "sre_geom_loss": 0.0021560368385834215,
      "sre_kd_loss": 8.689249644274824,
      "sre_logit_loss": 8.517362410052737,
      "sre_span_loss": 0.0640853275358677,
      "step": 6850,
      "vision_kd_loss": 1.04227843221277,
      "vision_logit_loss": 3.5169815029203892
    },
    {
      "affinity_loss": 0.006400096642028075,
      "emkd_kd_loss": 1.5192351310513914,
      "epoch": 0.5790017464454852,
      "grad_norm": 0.09216674417257309,
      "hard_loss": 0.8920129975369582,
      "joint_kd_loss": 4.378024808987975,
      "learning_rate": 3.973293344719407e-06,
      "loss": 2.6228411865234373,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 0.9734568951256756,
      "sre_geom_loss": 0.0020646166655568267,
      "sre_kd_loss": 7.236814487241208,
      "sre_logit_loss": 7.068428851420613,
      "sre_span_loss": 0.06515483496594243,
      "step": 6900,
      "vision_kd_loss": 1.0325066827982665,
      "vision_logit_loss": 3.4900170662999153
    },
    {
      "affinity_loss": 0.006469145946030039,
      "emkd_kd_loss": 1.5373933305218817,
      "epoch": 0.5831974112748003,
      "grad_norm": 0.16612648963928223,
      "hard_loss": 0.8797710109558011,
      "joint_kd_loss": 3.781782501153648,
      "learning_rate": 3.9068969516084e-06,
      "loss": 2.7946102905273436,
      "matched_vision_samples": 0.959375,
      "response_kd_loss": 0.9960211914334423,
      "sre_geom_loss": 0.00173324854933945,
      "sre_kd_loss": 6.026171702828724,
      "sre_logit_loss": 5.876540464416903,
      "sre_span_loss": 0.06296884158044122,
      "step": 6950,
      "vision_kd_loss": 1.0393827352672815,
      "vision_logit_loss": 3.5106163468956946
    },
    {
      "affinity_loss": 0.006509541520790663,
      "emkd_kd_loss": 1.5545906253438444,
      "epoch": 0.5873930761041154,
      "grad_norm": 0.15582379698753357,
      "hard_loss": 0.8430932255603488,
      "joint_kd_loss": 3.279548090957105,
      "learning_rate": 3.840702385146155e-06,
      "loss": 2.3933140563964845,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0260174973287042,
      "sre_geom_loss": 0.0021282103926552765,
      "sre_kd_loss": 5.004505573099014,
      "sre_logit_loss": 4.835104007148621,
      "sre_span_loss": 0.06299103119643405,
      "step": 7000,
      "vision_kd_loss": 1.0415818783268334,
      "vision_logit_loss": 3.5153733633458613
    },
    {
      "affinity_loss": 0.006525189042731654,
      "emkd_kd_loss": 1.5676670285686851,
      "epoch": 0.5915887409334305,
      "grad_norm": 0.4785439968109131,
      "hard_loss": 0.8951385973278866,
      "joint_kd_loss": 3.9708485272713006,
      "learning_rate": 3.7747218672613663e-06,
      "loss": 2.4379228210449218,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0461101430678081,
      "sre_geom_loss": 0.00197554460235736,
      "sre_kd_loss": 6.374029996783938,
      "sre_logit_loss": 6.2123023437328815,
      "sre_span_loss": 0.06295045091188513,
      "step": 7050,
      "vision_kd_loss": 1.0446119567379355,
      "vision_logit_loss": 3.525928916782141
    },
    {
      "affinity_loss": 0.006503448230068898,
      "emkd_kd_loss": 1.5520143486186861,
      "epoch": 0.5957844057627456,
      "grad_norm": 0.11826717853546143,
      "hard_loss": 0.8788637887573532,
      "joint_kd_loss": 4.824839720875024,
      "learning_rate": 3.708967580361565e-06,
      "loss": 2.569927062988281,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.0101261844916634,
      "sre_geom_loss": 0.0019096346366015383,
      "sre_kd_loss": 8.097665127050131,
      "sre_logit_loss": 7.9387908670520115,
      "sre_span_loss": 0.06339247009833343,
      "step": 7100,
      "vision_kd_loss": 1.046951256133616,
      "vision_logit_loss": 3.5374602022767068
    },
    {
      "affinity_loss": 0.006478808167885291,
      "emkd_kd_loss": 1.5158078995719553,
      "epoch": 0.5999800705920607,
      "grad_norm": 0.18883833289146423,
      "hard_loss": 0.8429051918849972,
      "joint_kd_loss": 4.2714817318506535,
      "learning_rate": 3.643451665083801e-06,
      "loss": 2.5203182983398436,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 0.9506644828793424,
      "sre_geom_loss": 0.0018533283664007526,
      "sre_kd_loss": 7.027155630479101,
      "sre_logit_loss": 6.87071371862141,
      "sre_span_loss": 0.06377550865639932,
      "step": 7150,
      "vision_kd_loss": 1.0404756585881114,
      "vision_logit_loss": 3.5140218202769757
    },
    {
      "affinity_loss": 0.006507578988821479,
      "emkd_kd_loss": 1.5331053482554853,
      "epoch": 0.6041757354213758,
      "grad_norm": 0.4692069888114929,
      "hard_loss": 0.8531817762082119,
      "joint_kd_loss": 4.408200336042792,
      "learning_rate": 3.5781862180530337e-06,
      "loss": 2.6625326538085936,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 0.9811332316247774,
      "sre_geom_loss": 0.002114735476353573,
      "sre_kd_loss": 7.283295310136164,
      "sre_logit_loss": 7.114334152443771,
      "sre_span_loss": 0.06322446241276339,
      "step": 7200,
      "vision_kd_loss": 1.04253873154521,
      "vision_logit_loss": 3.519397023320198
    },
    {
      "affinity_loss": 0.006397484335029731,
      "emkd_kd_loss": 1.5632163361366838,
      "epoch": 0.6083714002506909,
      "grad_norm": 0.06735660880804062,
      "hard_loss": 0.8598102957666334,
      "joint_kd_loss": 3.86629915792495,
      "learning_rate": 3.513183289648665e-06,
      "loss": 2.4306996154785154,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0671387001921153,
      "sre_geom_loss": 0.001972701095197884,
      "sre_kd_loss": 6.169382003282663,
      "sre_logit_loss": 6.008015555152233,
      "sre_span_loss": 0.0627313980134204,
      "step": 7250,
      "vision_kd_loss": 1.0296469866856932,
      "vision_logit_loss": 3.478839517980814
    },
    {
      "affinity_loss": 0.006433965254109353,
      "emkd_kd_loss": 1.5653763918951153,
      "epoch": 0.6125670650800061,
      "grad_norm": 0.3159357011318207,
      "hard_loss": 0.8719394440405267,
      "joint_kd_loss": 3.1388023188337684,
      "learning_rate": 3.448454881779584e-06,
      "loss": 2.4054531860351562,
      "matched_vision_samples": 0.953125,
      "response_kd_loss": 1.0633936626087586,
      "sre_geom_loss": 0.0019893507675638133,
      "sre_kd_loss": 4.712228244455764,
      "sre_logit_loss": 4.549556677418296,
      "sre_span_loss": 0.0632040416484233,
      "step": 7300,
      "vision_kd_loss": 1.033679560124874,
      "vision_logit_loss": 3.4913217206299305
    },
    {
      "affinity_loss": 0.006491441135149216,
      "emkd_kd_loss": 1.5518057198263704,
      "epoch": 0.6167627299093212,
      "grad_norm": 0.2627444863319397,
      "hard_loss": 0.9125513539720396,
      "joint_kd_loss": 4.197214244902134,
      "learning_rate": 3.3840129456681967e-06,
      "loss": 2.4607589721679686,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.035529186364522,
      "sre_geom_loss": 0.0019635451137304293,
      "sre_kd_loss": 6.8426227721211035,
      "sre_logit_loss": 6.6814030436002945,
      "sre_span_loss": 0.06304246791522018,
      "step": 7350,
      "vision_kd_loss": 1.034041124843061,
      "vision_logit_loss": 3.4870203916728495
    },
    {
      "affinity_loss": 0.006583787590061547,
      "emkd_kd_loss": 1.5509899496659636,
      "epoch": 0.6209583947386363,
      "grad_norm": 0.23648954927921295,
      "hard_loss": 0.8933240496744839,
      "joint_kd_loss": 4.838565495032817,
      "learning_rate": 3.31986937964378e-06,
      "loss": 2.61395751953125,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0108794568219435,
      "sre_geom_loss": 0.002061139555652858,
      "sre_kd_loss": 8.126141058312497,
      "sre_logit_loss": 7.958564820552201,
      "sre_span_loss": 0.06451924169203266,
      "step": 7400,
      "vision_kd_loss": 1.0455502200499176,
      "vision_logit_loss": 3.5238221211731435
    },
    {
      "affinity_loss": 0.00648064496461302,
      "emkd_kd_loss": 1.55110404712148,
      "epoch": 0.6251540595679514,
      "grad_norm": 0.6720110774040222,
      "hard_loss": 0.8698079175363091,
      "joint_kd_loss": 4.667755781058222,
      "learning_rate": 3.256036026945624e-06,
      "loss": 2.567772216796875,
      "matched_vision_samples": 0.958125,
      "response_kd_loss": 1.0222400400968286,
      "sre_geom_loss": 0.0019366300409185944,
      "sre_kd_loss": 7.784407572060591,
      "sre_logit_loss": 7.623950830210088,
      "sre_span_loss": 0.06362520858179778,
      "step": 7450,
      "vision_kd_loss": 1.039984029084444,
      "vision_logit_loss": 3.5118716223537922
    },
    {
      "affinity_loss": 0.006479496046813438,
      "emkd_kd_loss": 1.5433633477892728,
      "epoch": 0.6293497243972666,
      "grad_norm": 0.1313946545124054,
      "hard_loss": 0.8646195906128787,
      "joint_kd_loss": 4.552981555778533,
      "learning_rate": 3.1925246735363357e-06,
      "loss": 2.3972523498535154,
      "matched_vision_samples": 0.955625,
      "response_kd_loss": 1.0050711652267352,
      "sre_geom_loss": 0.0018978833187617995,
      "sre_kd_loss": 7.562599778238218,
      "sre_logit_loss": 7.404442032820661,
      "sre_span_loss": 0.06326358960941433,
      "step": 7500,
      "vision_kd_loss": 1.0408277658745646,
      "vision_logit_loss": 3.5153614611923696
    },
    {
      "affinity_loss": 0.006509763923677383,
      "emkd_kd_loss": 1.5603969538584352,
      "epoch": 0.6335453892265817,
      "grad_norm": 0.07378602027893066,
      "hard_loss": 0.8704146197348563,
      "joint_kd_loss": 4.477617089506238,
      "learning_rate": 3.1293470459257237e-06,
      "loss": 2.398359680175781,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 1.0395768175259787,
      "sre_geom_loss": 0.0020045735480479364,
      "sre_kd_loss": 7.39483725266764,
      "sre_logit_loss": 7.231303739409632,
      "sre_span_loss": 0.06330483971745707,
      "step": 7550,
      "vision_kd_loss": 1.0406085452437401,
      "vision_logit_loss": 3.5114577881991864
    },
    {
      "affinity_loss": 0.006380788386886707,
      "emkd_kd_loss": 1.5189769314508885,
      "epoch": 0.6377410540558968,
      "grad_norm": 0.10553375631570816,
      "hard_loss": 0.8649891205326253,
      "joint_kd_loss": 4.823087210059166,
      "learning_rate": 3.066514809005652e-06,
      "loss": 2.734862060546875,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 0.9858502069160306,
      "sre_geom_loss": 0.001808413770770385,
      "sre_kd_loss": 8.127197499151807,
      "sre_logit_loss": 7.97318782928749,
      "sre_span_loss": 0.06358897941070608,
      "step": 7600,
      "vision_kd_loss": 1.0260518277063966,
      "vision_logit_loss": 3.4661284686625002
    },
    {
      "affinity_loss": 0.006403710494050756,
      "emkd_kd_loss": 1.5028977679926903,
      "epoch": 0.6419367188852119,
      "grad_norm": 0.24485766887664795,
      "hard_loss": 0.8543473546673703,
      "joint_kd_loss": 3.7414077012054623,
      "learning_rate": 3.00403956389628e-06,
      "loss": 2.6687005615234374,
      "matched_vision_samples": 0.953125,
      "response_kd_loss": 0.9269612478624322,
      "sre_geom_loss": 0.0019604095634557697,
      "sre_kd_loss": 5.979917674423195,
      "sre_logit_loss": 5.818527316788532,
      "sre_span_loss": 0.06336989709991031,
      "step": 7650,
      "vision_kd_loss": 1.0394171449542045,
      "vision_logit_loss": 3.5172975340485575
    },
    {
      "affinity_loss": 0.0064617574265867,
      "emkd_kd_loss": 1.5643689604662359,
      "epoch": 0.646132383714527,
      "grad_norm": 0.07569508999586105,
      "hard_loss": 0.8467778444818305,
      "joint_kd_loss": 4.9329613002948465,
      "learning_rate": 2.941932845804074e-06,
      "loss": 2.735879211425781,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0416543260985953,
      "sre_geom_loss": 0.0019346856157610316,
      "sre_kd_loss": 8.301553567340598,
      "sre_logit_loss": 8.141888443622301,
      "sre_span_loss": 0.06293079920578748,
      "step": 7700,
      "vision_kd_loss": 1.0435417956486344,
      "vision_logit_loss": 3.5279914397001266
    },
    {
      "affinity_loss": 0.006499995580816176,
      "emkd_kd_loss": 1.5376164614362642,
      "epoch": 0.6503280485438421,
      "grad_norm": 0.26611676812171936,
      "hard_loss": 0.8899268394369801,
      "joint_kd_loss": 5.539626023285091,
      "learning_rate": 2.8802061218919846e-06,
      "loss": 2.833350830078125,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 0.996659777355701,
      "sre_geom_loss": 0.001952211186599726,
      "sre_kd_loss": 9.541635576132684,
      "sre_logit_loss": 9.380771290935227,
      "sre_span_loss": 0.06325365708093159,
      "step": 7750,
      "vision_kd_loss": 1.0392865712195636,
      "vision_logit_loss": 3.507146725654602
    },
    {
      "affinity_loss": 0.006502603554690722,
      "emkd_kd_loss": 1.542557702595368,
      "epoch": 0.6545237133731572,
      "grad_norm": 0.24941112101078033,
      "hard_loss": 0.8460189920223274,
      "joint_kd_loss": 4.084814104717225,
      "learning_rate": 2.8188707891621954e-06,
      "loss": 2.7140606689453124,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 0.9947674668951936,
      "sre_geom_loss": 0.0019547373502016453,
      "sre_kd_loss": 6.627070504061412,
      "sre_logit_loss": 6.465744993773406,
      "sre_span_loss": 0.063588631717721,
      "step": 7800,
      "vision_kd_loss": 1.04517397031188,
      "vision_logit_loss": 3.5304355199635027
    },
    {
      "affinity_loss": 0.00654169970744988,
      "emkd_kd_loss": 1.5385817897506058,
      "epoch": 0.6587193782024723,
      "grad_norm": 0.0741327628493309,
      "hard_loss": 0.8894039162669469,
      "joint_kd_loss": 4.524291731696576,
      "learning_rate": 2.757938172351808e-06,
      "loss": 2.5669244384765624,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0028540404523016,
      "sre_geom_loss": 0.00189080123445315,
      "sre_kd_loss": 7.510001678739209,
      "sre_logit_loss": 7.351196632216161,
      "sre_span_loss": 0.06426500045345165,
      "step": 7850,
      "vision_kd_loss": 1.0371547706052662,
      "vision_logit_loss": 3.4944491106271744
    },
    {
      "affinity_loss": 0.006461074958642712,
      "emkd_kd_loss": 1.5446503590419889,
      "epoch": 0.6629150430317874,
      "grad_norm": 0.18479380011558533,
      "hard_loss": 0.8865579040131343,
      "joint_kd_loss": 5.502392743602395,
      "learning_rate": 2.697419521841905e-06,
      "loss": 2.8456561279296877,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.025674361113986,
      "sre_geom_loss": 0.0020829868299927765,
      "sre_kd_loss": 9.460135121960192,
      "sre_logit_loss": 9.29124837052601,
      "sre_span_loss": 0.06473736370564438,
      "step": 7900,
      "vision_kd_loss": 1.0318131804838777,
      "vision_logit_loss": 3.4811452271044256
    },
    {
      "affinity_loss": 0.006449761293188203,
      "emkd_kd_loss": 1.5609519079420715,
      "epoch": 0.6671107078611025,
      "grad_norm": 0.11459030956029892,
      "hard_loss": 0.8416829336003139,
      "joint_kd_loss": 4.521407294347882,
      "learning_rate": 2.637326011580301e-06,
      "loss": 2.4996931457519533,
      "matched_vision_samples": 0.943125,
      "response_kd_loss": 1.0525440295918498,
      "sre_geom_loss": 0.0020204804914315754,
      "sre_kd_loss": 7.481862671412528,
      "sre_logit_loss": 7.317537122660724,
      "sre_span_loss": 0.06330153999035247,
      "step": 7950,
      "vision_kd_loss": 1.0346798949316145,
      "vision_logit_loss": 3.4937434519827364
    },
    {
      "affinity_loss": 0.006468181283271406,
      "emkd_kd_loss": 1.534418409699574,
      "epoch": 0.6713063726904176,
      "grad_norm": 0.5363131761550903,
      "hard_loss": 0.8399557377648762,
      "joint_kd_loss": 3.584399484284222,
      "learning_rate": 2.5776687370184326e-06,
      "loss": 2.3536083984375,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 0.9837722697181936,
      "sre_geom_loss": 0.0018703068746448181,
      "sre_kd_loss": 5.634380592105445,
      "sre_logit_loss": 5.479249491278461,
      "sre_span_loss": 0.06161572594777681,
      "step": 8000,
      "vision_kd_loss": 1.0425322763249278,
      "vision_logit_loss": 3.523310983926058
    },
    {
      "affinity_loss": 0.00641690119242412,
      "emkd_kd_loss": 1.5120388283859938,
      "epoch": 0.6755020375197327,
      "grad_norm": 0.255879670381546,
      "hard_loss": 0.8692767439931777,
      "joint_kd_loss": 4.777877583242953,
      "learning_rate": 2.518458713062732e-06,
      "loss": 2.6672879028320313,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 0.9595808976458529,
      "sre_geom_loss": 0.0018777655515892987,
      "sre_kd_loss": 8.043716268854915,
      "sre_logit_loss": 7.886146246501812,
      "sre_span_loss": 0.06368173877708613,
      "step": 8050,
      "vision_kd_loss": 1.0322483785450458,
      "vision_logit_loss": 3.4873033952713013
    },
    {
      "affinity_loss": 0.0064466234942665325,
      "emkd_kd_loss": 1.5137636885559187,
      "epoch": 0.6796977023490478,
      "grad_norm": 0.2170199602842331,
      "hard_loss": 0.8683581047238386,
      "joint_kd_loss": 4.8152068054303525,
      "learning_rate": 2.459706872040872e-06,
      "loss": 2.717254638671875,
      "matched_vision_samples": 0.9425,
      "response_kd_loss": 0.9591193172891178,
      "sre_geom_loss": 0.0019380608182000892,
      "sre_kd_loss": 8.116649915988091,
      "sre_logit_loss": 7.95503898931318,
      "sre_span_loss": 0.06470794893451966,
      "step": 8100,
      "vision_kd_loss": 1.0342040308564902,
      "vision_logit_loss": 3.492153775990009
    },
    {
      "affinity_loss": 0.006510985997447278,
      "emkd_kd_loss": 1.548429741319269,
      "epoch": 0.6838933671783629,
      "grad_norm": 0.21355466544628143,
      "hard_loss": 0.8631198805930217,
      "joint_kd_loss": 3.3585729659348726,
      "learning_rate": 2.4014240616832696e-06,
      "loss": 2.470103759765625,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 1.0051419106938704,
      "sre_geom_loss": 0.0018488861065829366,
      "sre_kd_loss": 5.168716207527322,
      "sre_logit_loss": 5.014362658774189,
      "sre_span_loss": 0.06190925201633945,
      "step": 8150,
      "vision_kd_loss": 1.0458587829768657,
      "vision_logit_loss": 3.532336531281471
    },
    {
      "affinity_loss": 0.0064477588961017315,
      "emkd_kd_loss": 1.565571507969871,
      "epoch": 0.688089032007678,
      "grad_norm": 0.10196200758218765,
      "hard_loss": 0.8353838110356719,
      "joint_kd_loss": 3.6114671663194895,
      "learning_rate": 2.343621043120186e-06,
      "loss": 2.595096435546875,
      "matched_vision_samples": 0.964375,
      "response_kd_loss": 1.0321422065620844,
      "sre_geom_loss": 0.0017952165405123565,
      "sre_kd_loss": 5.657362821607385,
      "sre_logit_loss": 5.505367870997434,
      "sre_span_loss": 0.06223413683357649,
      "step": 8200,
      "vision_kd_loss": 1.0495004061236977,
      "vision_logit_loss": 3.553225732445717
    },
    {
      "affinity_loss": 0.006625122145924252,
      "emkd_kd_loss": 1.5392022164538504,
      "epoch": 0.6922846968369932,
      "grad_norm": 0.507136344909668,
      "hard_loss": 0.8720541392471114,
      "joint_kd_loss": 3.7566976862587036,
      "learning_rate": 2.2863084888948385e-06,
      "loss": 2.4339138793945314,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 0.9812453545505775,
      "sre_geom_loss": 0.002111424394197421,
      "sre_kd_loss": 5.974193196288542,
      "sre_logit_loss": 5.804660800639685,
      "sre_span_loss": 0.06396118380129337,
      "step": 8250,
      "vision_kd_loss": 1.0485795386508108,
      "vision_logit_loss": 3.531805942058563
    },
    {
      "affinity_loss": 0.006386077318165917,
      "emkd_kd_loss": 1.5675778228603303,
      "epoch": 0.6964803616663083,
      "grad_norm": 0.2618618309497833,
      "hard_loss": 0.8939737452831378,
      "joint_kd_loss": 4.57378476569429,
      "learning_rate": 2.2294969809928578e-06,
      "loss": 2.5202423095703126,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.0954485928503026,
      "sre_geom_loss": 0.0020090420252034135,
      "sre_kd_loss": 7.5799916913639755,
      "sre_logit_loss": 7.414250395112176,
      "sre_span_loss": 0.06528919259086251,
      "step": 8300,
      "vision_kd_loss": 1.01985352601856,
      "vision_logit_loss": 3.4408063720166684
    },
    {
      "affinity_loss": 0.006467007260798709,
      "emkd_kd_loss": 1.5638732270523905,
      "epoch": 0.7006760264956234,
      "grad_norm": 0.19042706489562988,
      "hard_loss": 0.8519272471590511,
      "joint_kd_loss": 3.7960026844590904,
      "learning_rate": 2.1731970088884685e-06,
      "loss": 2.4526571655273437,
      "matched_vision_samples": 0.958125,
      "response_kd_loss": 1.0470514073937942,
      "sre_geom_loss": 0.0019182520028169848,
      "sre_kd_loss": 6.028132184809074,
      "sre_logit_loss": 5.868640788748453,
      "sre_span_loss": 0.06357875830726697,
      "step": 8350,
      "vision_kd_loss": 1.040347524136305,
      "vision_logit_loss": 3.514689370840788
    },
    {
      "affinity_loss": 0.0064372913949773646,
      "emkd_kd_loss": 1.512595064714551,
      "epoch": 0.7048716913249385,
      "grad_norm": 0.14714403450489044,
      "hard_loss": 0.840707841660278,
      "joint_kd_loss": 4.491656526513398,
      "learning_rate": 2.11741896760775e-06,
      "loss": 2.70340576171875,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 0.9459495308107512,
      "sre_geom_loss": 0.002010121190670411,
      "sre_kd_loss": 7.470717939126771,
      "sre_logit_loss": 7.306162350087834,
      "sre_span_loss": 0.06404955341131427,
      "step": 8400,
      "vision_kd_loss": 1.0396202990040182,
      "vision_logit_loss": 3.514752058684826
    },
    {
      "affinity_loss": 0.00645764164684806,
      "emkd_kd_loss": 1.5129334161058068,
      "epoch": 0.7090673561542536,
      "grad_norm": 0.15525202453136444,
      "hard_loss": 0.8789211065008806,
      "joint_kd_loss": 4.382952083814889,
      "learning_rate": 2.062173155809335e-06,
      "loss": 2.735565185546875,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 0.9583203494095165,
      "sre_geom_loss": 0.002111284272714613,
      "sre_kd_loss": 7.252970735190902,
      "sre_logit_loss": 7.083848291337199,
      "sre_span_loss": 0.06355817580129951,
      "step": 8450,
      "vision_kd_loss": 1.0337732421979309,
      "vision_logit_loss": 3.4893288005888463
    },
    {
      "affinity_loss": 0.00648659053753363,
      "emkd_kd_loss": 1.5448016021214426,
      "epoch": 0.7132630209835688,
      "grad_norm": 0.2253471463918686,
      "hard_loss": 0.8667475112096099,
      "joint_kd_loss": 4.348632848709822,
      "learning_rate": 2.007469773882901e-06,
      "loss": 2.55675537109375,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.0162817733415306,
      "sre_geom_loss": 0.001945880173610277,
      "sre_kd_loss": 7.15246412810171,
      "sre_logit_loss": 6.9908136926144655,
      "sre_span_loss": 0.064356399072567,
      "step": 8500,
      "vision_kd_loss": 1.036660713404417,
      "vision_logit_loss": 3.49798379778862
    },
    {
      "affinity_loss": 0.0063988302389043385,
      "emkd_kd_loss": 1.5320430529024451,
      "epoch": 0.7174586858128839,
      "grad_norm": 0.12183704227209091,
      "hard_loss": 0.8631034130768285,
      "joint_kd_loss": 4.022468880154192,
      "learning_rate": 1.9533189220658143e-06,
      "loss": 2.5648577880859373,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.0010481222645797,
      "sre_geom_loss": 0.0019221354454535345,
      "sre_kd_loss": 6.51289467350347,
      "sre_logit_loss": 6.35325263721752,
      "sre_span_loss": 0.06353524428675883,
      "step": 8550,
      "vision_kd_loss": 1.0315189919993282,
      "vision_logit_loss": 3.48619294449687
    },
    {
      "affinity_loss": 0.006514829481311608,
      "emkd_kd_loss": 1.5255341694131495,
      "epoch": 0.721654350642199,
      "grad_norm": 0.06661823391914368,
      "hard_loss": 0.8740396843025429,
      "joint_kd_loss": 4.612624001242221,
      "learning_rate": 1.8997305985782466e-06,
      "loss": 2.5010997009277345,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 0.987903381674306,
      "sre_geom_loss": 0.0018633666321331787,
      "sre_kd_loss": 7.699713821357582,
      "sre_logit_loss": 7.542663019531829,
      "sre_span_loss": 0.06388247242546641,
      "step": 8600,
      "vision_kd_loss": 1.031582477465272,
      "vision_logit_loss": 3.4748469580709935
    },
    {
      "affinity_loss": 0.006396955418749712,
      "emkd_kd_loss": 1.5166991803981364,
      "epoch": 0.7258500154715141,
      "grad_norm": 0.1973019391298294,
      "hard_loss": 0.8288065402424217,
      "joint_kd_loss": 4.588027210999281,
      "learning_rate": 1.8467146977771454e-06,
      "loss": 2.5716342163085937,
      "matched_vision_samples": 0.944375,
      "response_kd_loss": 0.9715667982591231,
      "sre_geom_loss": 0.001901741987966723,
      "sre_kd_loss": 7.659355222610757,
      "sre_logit_loss": 7.501294727001514,
      "sre_span_loss": 0.06297338944044896,
      "step": 8650,
      "vision_kd_loss": 1.0309157808870077,
      "vision_logit_loss": 3.483967584222555
    },
    {
      "affinity_loss": 0.00652289364268654,
      "emkd_kd_loss": 1.5367712777294218,
      "epoch": 0.7300456803008292,
      "grad_norm": 0.2566707134246826,
      "hard_loss": 0.8664904153991574,
      "joint_kd_loss": 4.098227983899415,
      "learning_rate": 1.7942810083293743e-06,
      "loss": 2.1676934814453124,
      "matched_vision_samples": 0.945625,
      "response_kd_loss": 1.0021667661879723,
      "sre_geom_loss": 0.001930164695466239,
      "sre_kd_loss": 6.659684714539908,
      "sre_logit_loss": 6.498557128591056,
      "sre_span_loss": 0.06461929590557702,
      "step": 8700,
      "vision_kd_loss": 1.0356878935173155,
      "vision_logit_loss": 3.490462208688259
    },
    {
      "affinity_loss": 0.006395021833741339,
      "emkd_kd_loss": 1.5383620795048774,
      "epoch": 0.7342413451301443,
      "grad_norm": 0.09541481733322144,
      "hard_loss": 0.8906733578299009,
      "joint_kd_loss": 3.532982423361391,
      "learning_rate": 1.7424392114043736e-06,
      "loss": 2.4720184326171877,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.0251957092488737,
      "sre_geom_loss": 0.002001666052834956,
      "sre_kd_loss": 5.527602786377538,
      "sre_logit_loss": 5.364457160625461,
      "sre_span_loss": 0.06306229146663099,
      "step": 8750,
      "vision_kd_loss": 1.0257642235234379,
      "vision_logit_loss": 3.4635547053813935
    },
    {
      "affinity_loss": 0.006688651167351054,
      "emkd_kd_loss": 1.5398693014495075,
      "epoch": 0.7384370099594594,
      "grad_norm": 0.09088502079248428,
      "hard_loss": 0.8451404463291419,
      "joint_kd_loss": 3.063192939534783,
      "learning_rate": 1.6911988788866563e-06,
      "loss": 2.502216339111328,
      "matched_vision_samples": 0.959375,
      "response_kd_loss": 0.9616113943606035,
      "sre_geom_loss": 0.0018644231471959926,
      "sre_kd_loss": 4.586516559522133,
      "sre_logit_loss": 4.430237426244203,
      "sre_span_loss": 0.06305797872017138,
      "step": 8800,
      "vision_kd_loss": 1.0590636022388935,
      "vision_logit_loss": 3.5673892915248873
    },
    {
      "affinity_loss": 0.0066219623161305205,
      "emkd_kd_loss": 1.573574535921216,
      "epoch": 0.7426326747887745,
      "grad_norm": 0.16084551811218262,
      "hard_loss": 0.8643264426286714,
      "joint_kd_loss": 3.2156070112250745,
      "learning_rate": 1.6405694716084864e-06,
      "loss": 2.3261984252929686,
      "matched_vision_samples": 0.956875,
      "response_kd_loss": 1.0646844617048554,
      "sre_geom_loss": 0.0019358498802930057,
      "sre_kd_loss": 4.857639470607974,
      "sre_logit_loss": 4.696402553949156,
      "sre_span_loss": 0.06444439831073395,
      "step": 8850,
      "vision_kd_loss": 1.0412323065847158,
      "vision_logit_loss": 3.5027329970896246
    },
    {
      "affinity_loss": 0.006538896976271644,
      "emkd_kd_loss": 1.5306817630119622,
      "epoch": 0.7468283396180896,
      "grad_norm": 0.08855391293764114,
      "hard_loss": 0.8538292177647144,
      "joint_kd_loss": 4.138508390411735,
      "learning_rate": 1.5905603376030759e-06,
      "loss": 2.4406288146972654,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 0.9800771342320264,
      "sre_geom_loss": 0.0019612961425607266,
      "sre_kd_loss": 6.74633503266843,
      "sre_logit_loss": 6.585014520932163,
      "sre_span_loss": 0.06325569117674604,
      "step": 8900,
      "vision_kd_loss": 1.0406431970372796,
      "vision_logit_loss": 3.508683088719845
    },
    {
      "affinity_loss": 0.006513400496914983,
      "emkd_kd_loss": 1.5506217689439654,
      "epoch": 0.7510240044474047,
      "grad_norm": 0.22840499877929688,
      "hard_loss": 0.8846565305338209,
      "joint_kd_loss": 4.334707346856594,
      "learning_rate": 1.5411807103785942e-06,
      "loss": 2.6970672607421875,
      "matched_vision_samples": 0.951875,
      "response_kd_loss": 1.0200387291334476,
      "sre_geom_loss": 0.0019493992418192364,
      "sre_kd_loss": 7.11879295398714,
      "sre_logit_loss": 6.957482305250596,
      "sre_span_loss": 0.06384067187667825,
      "step": 8950,
      "vision_kd_loss": 1.0406024039909243,
      "vision_logit_loss": 3.511069563627243
    },
    {
      "affinity_loss": 0.006489224478718825,
      "emkd_kd_loss": 1.5606571590527891,
      "epoch": 0.7552196692767198,
      "grad_norm": 0.24356971681118011,
      "hard_loss": 0.8823018762475613,
      "joint_kd_loss": 3.606493286583573,
      "learning_rate": 1.492439707213319e-06,
      "loss": 2.4421286010742187,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.0598687695742592,
      "sre_geom_loss": 0.0020230321461152333,
      "sre_kd_loss": 5.652329430587124,
      "sre_logit_loss": 5.486664176678168,
      "sre_span_loss": 0.06451365798711776,
      "step": 9000,
      "vision_kd_loss": 1.0307227735593916,
      "vision_logit_loss": 3.4739686465263366
    },
    {
      "affinity_loss": 0.0065204486607399305,
      "emkd_kd_loss": 1.510356516938191,
      "epoch": 0.7594153341060349,
      "grad_norm": 0.4253215193748474,
      "hard_loss": 0.84859300981827,
      "joint_kd_loss": 3.811448292857967,
      "learning_rate": 1.4443463274722597e-06,
      "loss": 2.3707736206054686,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 0.9368156878171743,
      "sre_geom_loss": 0.0020292763420931693,
      "sre_kd_loss": 6.11254007457057,
      "sre_logit_loss": 5.946377599222178,
      "sre_span_loss": 0.06469864117098041,
      "step": 9050,
      "vision_kd_loss": 1.0419486707821488,
      "vision_logit_loss": 3.515749818980694
    },
    {
      "affinity_loss": 0.006451333309960319,
      "emkd_kd_loss": 1.5347757527697832,
      "epoch": 0.76361099893535,
      "grad_norm": 0.13448835909366608,
      "hard_loss": 0.8433142518321551,
      "joint_kd_loss": 3.6487192047201096,
      "learning_rate": 1.3969094509455462e-06,
      "loss": 2.3921769714355468,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0003215859623311,
      "sre_geom_loss": 0.001874168740439397,
      "sre_kd_loss": 5.762662678181659,
      "sre_logit_loss": 5.605981359970901,
      "sre_span_loss": 0.06297290270100347,
      "step": 9100,
      "vision_kd_loss": 1.034614959396422,
      "vision_logit_loss": 3.4933265022933484
    },
    {
      "affinity_loss": 0.006510696367477067,
      "emkd_kd_loss": 1.5418532821163535,
      "epoch": 0.7678066637646651,
      "grad_norm": 0.12729620933532715,
      "hard_loss": 0.8636211203911262,
      "joint_kd_loss": 3.8579187249578535,
      "learning_rate": 1.3501378362088991e-06,
      "loss": 2.477095947265625,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.015215746180969,
      "sre_geom_loss": 0.0019597004403079945,
      "sre_kd_loss": 6.17398419520352,
      "sre_logit_loss": 6.012575114659849,
      "sre_span_loss": 0.06342404543422163,
      "step": 9150,
      "vision_kd_loss": 1.034245410002768,
      "vision_logit_loss": 3.485912002623081
    },
    {
      "affinity_loss": 0.006410567889688536,
      "emkd_kd_loss": 1.5403886809479446,
      "epoch": 0.7720023285939803,
      "grad_norm": 0.18086719512939453,
      "hard_loss": 0.8830904738350819,
      "joint_kd_loss": 5.019010835550725,
      "learning_rate": 1.30404011900647e-06,
      "loss": 2.6548580932617187,
      "matched_vision_samples": 0.943125,
      "response_kd_loss": 1.0325805203789553,
      "sre_geom_loss": 0.0019443875596440918,
      "sre_kd_loss": 8.49763301921077,
      "sre_logit_loss": 8.335409147832543,
      "sre_span_loss": 0.06500449909828604,
      "step": 9200,
      "vision_kd_loss": 1.0240984203666448,
      "vision_logit_loss": 3.455336894392967
    },
    {
      "affinity_loss": 0.00649213824653998,
      "emkd_kd_loss": 1.5650797839649022,
      "epoch": 0.7761979934232954,
      "grad_norm": 0.13584552705287933,
      "hard_loss": 0.8702037201027725,
      "joint_kd_loss": 4.214695109575987,
      "learning_rate": 1.258624810656376e-06,
      "loss": 2.4308009338378906,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0527886766126164,
      "sre_geom_loss": 0.0018756390541447131,
      "sre_kd_loss": 6.864310459911358,
      "sre_logit_loss": 6.7068128782257554,
      "sre_span_loss": 0.06371559924678877,
      "step": 9250,
      "vision_kd_loss": 1.0386854458972812,
      "vision_logit_loss": 3.5055279597640037
    },
    {
      "affinity_loss": 0.006543343347439077,
      "emkd_kd_loss": 1.5498118914291263,
      "epoch": 0.7803936582526105,
      "grad_norm": 0.13774774968624115,
      "hard_loss": 0.8720281066143843,
      "joint_kd_loss": 4.501363873146475,
      "learning_rate": 1.2139002964791957e-06,
      "loss": 2.566243896484375,
      "matched_vision_samples": 0.948125,
      "response_kd_loss": 1.0351016355073461,
      "sre_geom_loss": 0.0020146517252615757,
      "sre_kd_loss": 7.452915849355049,
      "sre_logit_loss": 7.288031789141023,
      "sre_span_loss": 0.06415148838190361,
      "step": 9300,
      "vision_kd_loss": 1.0322610747814178,
      "vision_logit_loss": 3.4747099655866625
    },
    {
      "affinity_loss": 0.006583238494349644,
      "emkd_kd_loss": 1.5163593290001154,
      "epoch": 0.7845893230819256,
      "grad_norm": 0.06226549297571182,
      "hard_loss": 0.8438626476651734,
      "joint_kd_loss": 3.8512483131699264,
      "learning_rate": 1.1698748342497352e-06,
      "loss": 2.4173208618164064,
      "matched_vision_samples": 0.96,
      "response_kd_loss": 0.9414971954152258,
      "sre_geom_loss": 0.0017639147475936668,
      "sre_kd_loss": 6.1861372944782484,
      "sre_logit_loss": 6.0352069946708795,
      "sre_span_loss": 0.06273457553121262,
      "step": 9350,
      "vision_kd_loss": 1.0456107299774886,
      "vision_logit_loss": 3.524119074642658
    },
    {
      "affinity_loss": 0.006513422731222818,
      "emkd_kd_loss": 1.5605884413421154,
      "epoch": 0.7887849879112407,
      "grad_norm": 0.42850354313850403,
      "hard_loss": 0.8816883279940407,
      "joint_kd_loss": 3.193368703331798,
      "learning_rate": 1.1265565526723427e-06,
      "loss": 2.319779968261719,
      "matched_vision_samples": 0.96125,
      "response_kd_loss": 1.0431640561132405,
      "sre_geom_loss": 0.0019102828832889697,
      "sre_kd_loss": 4.826148915259401,
      "sre_logit_loss": 4.6684094816409925,
      "sre_span_loss": 0.06222530158935115,
      "step": 9400,
      "vision_kd_loss": 1.0390064128115772,
      "vision_logit_loss": 3.504683380126953
    },
    {
      "affinity_loss": 0.006485201576579129,
      "emkd_kd_loss": 1.5533982539270073,
      "epoch": 0.7929806527405558,
      "grad_norm": 0.19703303277492523,
      "hard_loss": 0.8795610651959191,
      "joint_kd_loss": 3.643765701260418,
      "learning_rate": 1.0839534498800525e-06,
      "loss": 2.6436236572265623,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 1.0441728729796766,
      "sre_geom_loss": 0.002014095123287092,
      "sre_kd_loss": 5.734133134281729,
      "sre_logit_loss": 5.56942435634046,
      "sre_span_loss": 0.06400399218895472,
      "step": 9450,
      "vision_kd_loss": 1.031311817318201,
      "vision_logit_loss": 3.4767271074652673
    },
    {
      "affinity_loss": 0.006634368442901177,
      "emkd_kd_loss": 1.53978007491678,
      "epoch": 0.797176317569871,
      "grad_norm": 0.10026870667934418,
      "hard_loss": 0.8628344030156131,
      "joint_kd_loss": 3.455086545627564,
      "learning_rate": 1.0420733919578351e-06,
      "loss": 2.388077697753906,
      "matched_vision_samples": 0.9625,
      "response_kd_loss": 0.9779665456648547,
      "sre_geom_loss": 0.0020272736432982085,
      "sre_kd_loss": 5.370392961505567,
      "sre_logit_loss": 5.205212472757121,
      "sre_span_loss": 0.06381679158424959,
      "step": 9500,
      "vision_kd_loss": 1.0507968020811678,
      "vision_logit_loss": 3.5397503603994847
    },
    {
      "affinity_loss": 0.006594160190288676,
      "emkd_kd_loss": 1.5348435321450233,
      "epoch": 0.8013719823991861,
      "grad_norm": 0.1634082943201065,
      "hard_loss": 0.8616184356800295,
      "joint_kd_loss": 5.951733742598444,
      "learning_rate": 1.0009241114902425e-06,
      "loss": 2.6169659423828127,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 0.9729060832291668,
      "sre_geom_loss": 0.0019319235117537304,
      "sre_kd_loss": 10.368623992280336,
      "sre_logit_loss": 10.20856187656871,
      "sre_span_loss": 0.06346594358561561,
      "step": 9550,
      "vision_kd_loss": 1.048390490449965,
      "vision_logit_loss": 3.5341459426283834
    },
    {
      "affinity_loss": 0.0064054963187663815,
      "emkd_kd_loss": 1.5283261058107018,
      "epoch": 0.8055676472285012,
      "grad_norm": 0.23133595287799835,
      "hard_loss": 0.8974019404403316,
      "joint_kd_loss": 4.226255514994263,
      "learning_rate": 9.605132061336758e-07,
      "loss": 2.6646002197265624,
      "matched_vision_samples": 0.94625,
      "response_kd_loss": 1.0211195779925832,
      "sre_geom_loss": 0.0020138616955467726,
      "sre_kd_loss": 6.924184936755337,
      "sre_logit_loss": 6.759561293619044,
      "sre_span_loss": 0.06393055396620184,
      "step": 9600,
      "vision_kd_loss": 1.017766316756606,
      "vision_logit_loss": 3.4305156360566618
    },
    {
      "affinity_loss": 0.006643425234942697,
      "emkd_kd_loss": 1.5449046015925705,
      "epoch": 0.8097633120578163,
      "grad_norm": 0.09143870323896408,
      "hard_loss": 0.8624225535498863,
      "joint_kd_loss": 3.2870538279600443,
      "learning_rate": 9.208481372135914e-07,
      "loss": 2.311671142578125,
      "matched_vision_samples": 0.95875,
      "response_kd_loss": 0.9977017307498898,
      "sre_geom_loss": 0.0019446711927911054,
      "sre_kd_loss": 5.029203023004811,
      "sre_logit_loss": 4.868618754749768,
      "sre_span_loss": 0.06335069384309464,
      "step": 9650,
      "vision_kd_loss": 1.0460537362471223,
      "vision_logit_loss": 3.519872421622276
    },
    {
      "affinity_loss": 0.006463352774881059,
      "emkd_kd_loss": 1.5392785628698766,
      "epoch": 0.8139589768871314,
      "grad_norm": 0.10640463978052139,
      "hard_loss": 0.8570492776481933,
      "joint_kd_loss": 4.879935857560485,
      "learning_rate": 8.819362283468663e-07,
      "loss": 2.5305410766601564,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0058014160167028,
      "sre_geom_loss": 0.0019064096297612388,
      "sre_kd_loss": 8.220593147273176,
      "sre_logit_loss": 8.062349847880977,
      "sre_span_loss": 0.06292281305650249,
      "step": 9700,
      "vision_kd_loss": 1.0363778549805283,
      "vision_logit_loss": 3.4991761435568334
    },
    {
      "affinity_loss": 0.006473256551689701,
      "emkd_kd_loss": 1.5275380058214068,
      "epoch": 0.8181546417164465,
      "grad_norm": 0.22732943296432495,
      "hard_loss": 0.8726426401726712,
      "joint_kd_loss": 4.040941182039678,
      "learning_rate": 8.437846640895896e-07,
      "loss": 2.505040740966797,
      "matched_vision_samples": 0.94875,
      "response_kd_loss": 0.98716237024174,
      "sre_geom_loss": 0.001982523817943758,
      "sre_kd_loss": 6.55434432410635,
      "sre_logit_loss": 6.391635249307001,
      "sre_span_loss": 0.06358285245951265,
      "step": 9750,
      "vision_kd_loss": 1.0339568216353654,
      "vision_logit_loss": 3.488501630872488
    },
    {
      "affinity_loss": 0.006535206022963393,
      "emkd_kd_loss": 1.5795160097442567,
      "epoch": 0.8223503065457616,
      "grad_norm": 0.3167291283607483,
      "hard_loss": 0.8678521314505361,
      "joint_kd_loss": 4.34757536213845,
      "learning_rate": 8.064004886105365e-07,
      "loss": 2.7270074462890626,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 1.0779974599214206,
      "sre_geom_loss": 0.001892072609069313,
      "sre_kd_loss": 7.115634711273015,
      "sre_logit_loss": 6.957914061278352,
      "sre_span_loss": 0.06311700082966126,
      "step": 9800,
      "vision_kd_loss": 1.0405172807723284,
      "vision_logit_loss": 3.508548520952463
    },
    {
      "affinity_loss": 0.006577427046577214,
      "emkd_kd_loss": 1.558403942277655,
      "epoch": 0.8265459713750767,
      "grad_norm": 0.2567388713359833,
      "hard_loss": 0.8670905783223497,
      "joint_kd_loss": 4.9212745535001154,
      "learning_rate": 7.697906043905518e-07,
      "loss": 2.4826759338378905,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 1.0514320740770091,
      "sre_geom_loss": 0.0018643595134637536,
      "sre_kd_loss": 8.284145144401117,
      "sre_logit_loss": 8.125809059062304,
      "sre_span_loss": 0.06511808631825261,
      "step": 9850,
      "vision_kd_loss": 1.0326879070326687,
      "vision_logit_loss": 3.473008925020695
    },
    {
      "affinity_loss": 0.006474300011759624,
      "emkd_kd_loss": 1.5610076847486198,
      "epoch": 0.8307416362043918,
      "grad_norm": 0.17452222108840942,
      "hard_loss": 0.9059534767829382,
      "joint_kd_loss": 4.631926656421274,
      "learning_rate": 7.339617709481028e-07,
      "loss": 2.6316241455078124,
      "matched_vision_samples": 0.945,
      "response_kd_loss": 1.0595079555893756,
      "sre_geom_loss": 0.002134101902919068,
      "sre_kd_loss": 7.7028455976920664,
      "sre_logit_loss": 7.531634335807903,
      "sre_span_loss": 0.06450617875088938,
      "step": 9900,
      "vision_kd_loss": 1.031253707781434,
      "vision_logit_loss": 3.477584832459688
    },
    {
      "affinity_loss": 0.0064533528810716235,
      "emkd_kd_loss": 1.5515110437292605,
      "epoch": 0.8349373010337069,
      "grad_norm": 0.20352788269519806,
      "hard_loss": 0.8882148785787103,
      "joint_kd_loss": 4.543164052478969,
      "learning_rate": 6.989206035912272e-07,
      "loss": 2.5219186401367186,
      "matched_vision_samples": 0.946875,
      "response_kd_loss": 1.0436449058588027,
      "sre_geom_loss": 0.0019639114617485643,
      "sre_kd_loss": 7.53481711694505,
      "sre_logit_loss": 7.373098492567369,
      "sre_span_loss": 0.06352309837588109,
      "step": 9950,
      "vision_kd_loss": 1.029688589759171,
      "vision_logit_loss": 3.4734190756082537
    },
    {
      "affinity_loss": 0.006488749169075163,
      "emkd_kd_loss": 1.541732177408412,
      "epoch": 0.839132965863022,
      "grad_norm": 0.06896962970495224,
      "hard_loss": 0.8537093441864773,
      "joint_kd_loss": 4.335891288276762,
      "learning_rate": 6.646735721961078e-07,
      "loss": 2.4135169982910156,
      "matched_vision_samples": 0.948125,
      "response_kd_loss": 1.0062524174767253,
      "sre_geom_loss": 0.0018112153753940597,
      "sre_kd_loss": 7.13005045969272,
      "sre_logit_loss": 6.975819183430722,
      "sre_span_loss": 0.06367050549713894,
      "step": 10000,
      "vision_kd_loss": 1.0386059683933855,
      "vision_logit_loss": 3.5055489563941955
    },
    {
      "affinity_loss": 0.0065729232598096135,
      "emkd_kd_loss": 1.561451576575637,
      "epoch": 0.8433286306923371,
      "grad_norm": 0.2623848617076874,
      "hard_loss": 0.8722002245327712,
      "joint_kd_loss": 3.4214030976593492,
      "learning_rate": 6.312270000124937e-07,
      "loss": 2.3796807861328126,
      "matched_vision_samples": 0.956875,
      "response_kd_loss": 1.0384164511977065,
      "sre_geom_loss": 0.0019678929252446055,
      "sre_kd_loss": 5.281354611833812,
      "sre_logit_loss": 5.119999673432467,
      "sre_span_loss": 0.06296029683318921,
      "step": 10050,
      "vision_kd_loss": 1.0422433495521546,
      "vision_logit_loss": 3.5116810750961305
    },
    {
      "affinity_loss": 0.0064869433343119455,
      "emkd_kd_loss": 1.5406995825003833,
      "epoch": 0.8475242955216522,
      "grad_norm": 0.11369244754314423,
      "hard_loss": 0.884911906704474,
      "joint_kd_loss": 4.471731528751552,
      "learning_rate": 5.985870624961993e-07,
      "loss": 2.6841021728515626,
      "matched_vision_samples": 0.951875,
      "response_kd_loss": 1.012402890038129,
      "sre_geom_loss": 0.0019412229031522798,
      "sre_kd_loss": 7.4027634482341815,
      "sre_logit_loss": 7.2421040086253194,
      "sre_span_loss": 0.06359828345826828,
      "step": 10100,
      "vision_kd_loss": 1.0344981360435486,
      "vision_logit_loss": 3.489298209697008
    },
    {
      "affinity_loss": 0.0064956437291402835,
      "emkd_kd_loss": 1.51087097203359,
      "epoch": 0.8517199603509673,
      "grad_norm": 0.11900119483470917,
      "hard_loss": 0.8460070198279663,
      "joint_kd_loss": 3.9350012897700073,
      "learning_rate": 5.6675978616889e-07,
      "loss": 2.382200012207031,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 0.954871941176753,
      "sre_geom_loss": 0.0018794824685983258,
      "sre_kd_loss": 6.359131597196683,
      "sre_logit_loss": 6.202095793952903,
      "sre_span_loss": 0.06306164121371693,
      "step": 10150,
      "vision_kd_loss": 1.0334350034594535,
      "vision_logit_loss": 3.48417563945055
    },
    {
      "affinity_loss": 0.006454206977650756,
      "emkd_kd_loss": 1.5546030168700964,
      "epoch": 0.8559156251802825,
      "grad_norm": 0.10019361972808838,
      "hard_loss": 0.864901004832343,
      "joint_kd_loss": 3.5351383159682155,
      "learning_rate": 5.357510475053701e-07,
      "loss": 2.5337989807128904,
      "matched_vision_samples": 0.958125,
      "response_kd_loss": 1.0431655679121183,
      "sre_geom_loss": 0.001972841378218533,
      "sre_kd_loss": 5.515673627764918,
      "sre_logit_loss": 5.354439398128452,
      "sre_span_loss": 0.06259213362820447,
      "step": 10200,
      "vision_kd_loss": 1.0330202332139016,
      "vision_logit_loss": 3.4866602343320845
    },
    {
      "affinity_loss": 0.006475165198935429,
      "emkd_kd_loss": 1.5483875653147698,
      "epoch": 0.8601112900095976,
      "grad_norm": 0.1342584788799286,
      "hard_loss": 0.8682425983360372,
      "joint_kd_loss": 3.8989097480103374,
      "learning_rate": 5.055665718485653e-07,
      "loss": 2.4564215087890626,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.0194377220480715,
      "sre_geom_loss": 0.0019105496519665622,
      "sre_kd_loss": 6.249431961500086,
      "sre_logit_loss": 6.091122649396857,
      "sre_span_loss": 0.06278181096306071,
      "step": 10250,
      "vision_kd_loss": 1.0386687035113573,
      "vision_logit_loss": 3.507158299088478
    },
    {
      "affinity_loss": 0.006458835987432394,
      "emkd_kd_loss": 1.540869635362178,
      "epoch": 0.8643069548389127,
      "grad_norm": 1.1370502710342407,
      "hard_loss": 0.8355148534498631,
      "joint_kd_loss": 3.6291948861815033,
      "learning_rate": 4.762119323524217e-07,
      "loss": 2.65744384765625,
      "matched_vision_samples": 0.965,
      "response_kd_loss": 1.0166295015015845,
      "sre_geom_loss": 0.0019213729950030524,
      "sre_kd_loss": 5.717520124044968,
      "sre_logit_loss": 5.55964068257148,
      "sre_span_loss": 0.061810764932306486,
      "step": 10300,
      "vision_kd_loss": 1.0325548845157027,
      "vision_logit_loss": 3.484335938543081
    },
    {
      "affinity_loss": 0.006577316216717009,
      "emkd_kd_loss": 1.5520580302365123,
      "epoch": 0.8685026196682278,
      "grad_norm": 0.05618914216756821,
      "hard_loss": 0.8652371595972635,
      "joint_kd_loss": 4.26237755490467,
      "learning_rate": 4.4769254895289384e-07,
      "loss": 2.6089691162109374,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 1.0148884061996022,
      "sre_geom_loss": 0.0019511432251572546,
      "sre_kd_loss": 6.9726971078501085,
      "sre_logit_loss": 6.811356577342667,
      "sre_span_loss": 0.06378335204557516,
      "step": 10350,
      "vision_kd_loss": 1.0446138247847556,
      "vision_logit_loss": 3.520723679959774
    },
    {
      "affinity_loss": 0.006474891065008706,
      "emkd_kd_loss": 1.544711274653673,
      "epoch": 0.8726982844975429,
      "grad_norm": 0.15860378742218018,
      "hard_loss": 0.8847281838749496,
      "joint_kd_loss": 4.406491039842368,
      "learning_rate": 4.200136873672306e-07,
      "loss": 2.672891845703125,
      "matched_vision_samples": 0.9475,
      "response_kd_loss": 1.016786475962141,
      "sre_geom_loss": 0.0019959304070227193,
      "sre_kd_loss": 7.268270809906535,
      "sre_logit_loss": 7.1040767960315865,
      "sre_span_loss": 0.06439748642384074,
      "step": 10400,
      "vision_kd_loss": 1.0363180354982615,
      "vision_logit_loss": 3.497783037573099
    },
    {
      "affinity_loss": 0.006612303897854872,
      "emkd_kd_loss": 1.5604409501329064,
      "epoch": 0.876893949326858,
      "grad_norm": 0.10128392279148102,
      "hard_loss": 0.8671850239741616,
      "joint_kd_loss": 4.003001125603914,
      "learning_rate": 3.9318045812172964e-07,
      "loss": 2.7672528076171874,
      "matched_vision_samples": 0.956875,
      "response_kd_loss": 1.0247295266604612,
      "sre_geom_loss": 0.0019882123429747576,
      "sre_kd_loss": 6.445561308055185,
      "sre_logit_loss": 6.282943901293256,
      "sre_span_loss": 0.06320678067626431,
      "step": 10450,
      "vision_kd_loss": 1.0480761886760592,
      "vision_logit_loss": 3.531074366122484
    },
    {
      "affinity_loss": 0.0065333780227229,
      "emkd_kd_loss": 1.5299394098017365,
      "epoch": 0.8810896141561732,
      "grad_norm": 1.0709534883499146,
      "hard_loss": 0.8806658451482872,
      "joint_kd_loss": 4.824106566682458,
      "learning_rate": 3.671978156081507e-07,
      "loss": 2.576925048828125,
      "matched_vision_samples": 0.95125,
      "response_kd_loss": 0.980048872522051,
      "sre_geom_loss": 0.0020281816974446086,
      "sre_kd_loss": 8.118273752939421,
      "sre_logit_loss": 7.951748686076026,
      "sre_span_loss": 0.06511603475548328,
      "step": 10500,
      "vision_kd_loss": 1.0399149744585157,
      "vision_logit_loss": 3.506322091817856
    },
    {
      "affinity_loss": 0.006679793746297946,
      "emkd_kd_loss": 1.55503047573613,
      "epoch": 0.8852852789854883,
      "grad_norm": 0.10078469663858414,
      "hard_loss": 0.8980053924857293,
      "joint_kd_loss": 3.754646478034556,
      "learning_rate": 3.4207055716894866e-07,
      "loss": 2.1478826904296877,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.0188061803379163,
      "sre_geom_loss": 0.0020469615824871834,
      "sre_kd_loss": 5.954262515858281,
      "sre_logit_loss": 5.787868354953826,
      "sre_span_loss": 0.06404610192636027,
      "step": 10550,
      "vision_kd_loss": 1.0456273869797588,
      "vision_logit_loss": 3.514530173689127
    },
    {
      "affinity_loss": 0.006503608817438362,
      "emkd_kd_loss": 1.5878790758736432,
      "epoch": 0.8894809438148034,
      "grad_norm": 0.4025152027606964,
      "hard_loss": 0.8442065017181676,
      "joint_kd_loss": 4.0537421585805715,
      "learning_rate": 3.17803322211519e-07,
      "loss": 2.47606689453125,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.0908815696632963,
      "sre_geom_loss": 0.0019976356914427028,
      "sre_kd_loss": 6.5196052673342635,
      "sre_logit_loss": 6.357545089257473,
      "sre_span_loss": 0.06217837560107,
      "step": 10600,
      "vision_kd_loss": 1.0424382898584008,
      "vision_logit_loss": 3.5193922778964044
    },
    {
      "affinity_loss": 0.006454971600032877,
      "emkd_kd_loss": 1.517277034325525,
      "epoch": 0.8936766086441185,
      "grad_norm": 0.18001419305801392,
      "hard_loss": 0.892456656248969,
      "joint_kd_loss": 4.222473559975624,
      "learning_rate": 2.9440059135158407e-07,
      "loss": 2.5034439086914064,
      "matched_vision_samples": 0.9575,
      "response_kd_loss": 0.9818980109245069,
      "sre_geom_loss": 0.001804596533628633,
      "sre_kd_loss": 6.927670066184365,
      "sre_logit_loss": 6.774329802715511,
      "sre_span_loss": 0.06311046238290147,
      "step": 10650,
      "vision_kd_loss": 1.0263280281424523,
      "vision_logit_loss": 3.4598149512708187
    },
    {
      "affinity_loss": 0.006462680320837535,
      "emkd_kd_loss": 1.523248645234853,
      "epoch": 0.8978722734734336,
      "grad_norm": 0.16415078938007355,
      "hard_loss": 0.8741292747463604,
      "joint_kd_loss": 4.455010732058436,
      "learning_rate": 2.7186668558591313e-07,
      "loss": 2.721918029785156,
      "matched_vision_samples": 0.95625,
      "response_kd_loss": 0.9800283841221961,
      "sre_geom_loss": 0.0018206017757915304,
      "sre_kd_loss": 7.386772871774156,
      "sre_logit_loss": 7.232528321425198,
      "sre_span_loss": 0.06321439311257564,
      "step": 10700,
      "vision_kd_loss": 1.0332344532757998,
      "vision_logit_loss": 3.4866697813570497
    },
    {
      "affinity_loss": 0.00649449909789837,
      "emkd_kd_loss": 1.5716129031963646,
      "epoch": 0.9020679383027487,
      "grad_norm": 0.10341391712427139,
      "hard_loss": 0.8951796746472883,
      "joint_kd_loss": 5.234054196886718,
      "learning_rate": 2.5020576549451e-07,
      "loss": 2.7688339233398436,
      "matched_vision_samples": 0.95,
      "response_kd_loss": 1.086263558607667,
      "sre_geom_loss": 0.0019881837222407482,
      "sre_kd_loss": 8.896495470046066,
      "sre_logit_loss": 8.73339246520729,
      "sre_span_loss": 0.06369387886254117,
      "step": 10750,
      "vision_kd_loss": 1.028481121994555,
      "vision_logit_loss": 3.464474579691887
    },
    {
      "affinity_loss": 0.006621992945147212,
      "emkd_kd_loss": 1.543804144989699,
      "epoch": 0.9062636031320638,
      "grad_norm": 0.07538052648305893,
      "hard_loss": 0.8812437724468555,
      "joint_kd_loss": 3.07537722280249,
      "learning_rate": 2.294218304724144e-07,
      "loss": 2.2016250610351564,
      "matched_vision_samples": 0.963125,
      "response_kd_loss": 0.9900864872521623,
      "sre_geom_loss": 0.0018629465883011687,
      "sre_kd_loss": 4.606950291146059,
      "sre_logit_loss": 4.451733019700914,
      "sre_span_loss": 0.06206992555875331,
      "step": 10800,
      "vision_kd_loss": 1.0487609005719423,
      "vision_logit_loss": 3.5328443060815333
    },
    {
      "affinity_loss": 0.006613701859168941,
      "emkd_kd_loss": 1.5891315945517273,
      "epoch": 0.9104592679613789,
      "grad_norm": 0.1905500888824463,
      "hard_loss": 0.9072908033765088,
      "joint_kd_loss": 4.897401753999293,
      "learning_rate": 2.0951871799127333e-07,
      "loss": 2.5575827026367186,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.0898592409431536,
      "sre_geom_loss": 0.0020365191204515655,
      "sre_kd_loss": 8.20567188117886,
      "sre_logit_loss": 8.040095640060244,
      "sre_span_loss": 0.06375031703151762,
      "step": 10850,
      "vision_kd_loss": 1.0442019730806351,
      "vision_logit_loss": 3.515437706410885
    },
    {
      "affinity_loss": 0.006568835007637972,
      "emkd_kd_loss": 1.5710950529202818,
      "epoch": 0.914654932790694,
      "grad_norm": 0.1823987066745758,
      "hard_loss": 0.8811023505210688,
      "joint_kd_loss": 4.504920217711478,
      "learning_rate": 1.9050010289079934e-07,
      "loss": 2.583563232421875,
      "matched_vision_samples": 0.956875,
      "response_kd_loss": 1.0541294955184453,
      "sre_geom_loss": 0.001947018485320342,
      "sre_kd_loss": 7.438745355097344,
      "sre_logit_loss": 7.278654913412611,
      "sre_span_loss": 0.06273951887851581,
      "step": 10900,
      "vision_kd_loss": 1.0440303061902523,
      "vision_logit_loss": 3.51923772200942
    },
    {
      "affinity_loss": 0.006476775719056605,
      "emkd_kd_loss": 1.562457065936178,
      "epoch": 0.9188505976200091,
      "grad_norm": 0.16165553033351898,
      "hard_loss": 0.8924449981437738,
      "joint_kd_loss": 4.302009254060685,
      "learning_rate": 1.7236949670026037e-07,
      "loss": 2.3342385864257813,
      "matched_vision_samples": 0.953125,
      "response_kd_loss": 1.0714453314531829,
      "sre_geom_loss": 0.002002668391495774,
      "sre_kd_loss": 7.041561471216847,
      "sre_logit_loss": 6.87723534073215,
      "sre_span_loss": 0.0641927521082107,
      "step": 10950,
      "vision_kd_loss": 1.0267344009876251,
      "vision_logit_loss": 3.459260034561157
    },
    {
      "affinity_loss": 0.006610855693434132,
      "emkd_kd_loss": 1.5318699948163703,
      "epoch": 0.9230462624493242,
      "grad_norm": 0.23488570749759674,
      "hard_loss": 0.8693321059104665,
      "joint_kd_loss": 3.6864225023798642,
      "learning_rate": 1.5513024699013e-07,
      "loss": 2.2951785278320314,
      "matched_vision_samples": 0.951875,
      "response_kd_loss": 0.9812713352102287,
      "sre_geom_loss": 0.001870295328526268,
      "sre_kd_loss": 5.840974995878059,
      "sre_logit_loss": 5.684108877738763,
      "sre_span_loss": 0.06335139591712505,
      "step": 11000,
      "vision_kd_loss": 1.0412343273684384,
      "vision_logit_loss": 3.5038517402112483
    },
    {
      "affinity_loss": 0.006638653315312695,
      "emkd_kd_loss": 1.513115250794217,
      "epoch": 0.9272419272786393,
      "grad_norm": 0.3757422864437103,
      "hard_loss": 0.8721481731026915,
      "joint_kd_loss": 3.5608858956769107,
      "learning_rate": 1.3878553675399807e-07,
      "loss": 2.5274368286132813,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 0.9349886089653648,
      "sre_geom_loss": 0.0017475168884203108,
      "sre_kd_loss": 5.608656519184588,
      "sre_logit_loss": 5.458051718250354,
      "sre_span_loss": 0.06322894725133665,
      "step": 11050,
      "vision_kd_loss": 1.045620946995914,
      "vision_logit_loss": 3.5186184580624102
    },
    {
      "affinity_loss": 0.006562280898651807,
      "emkd_kd_loss": 1.5192983460519462,
      "epoch": 0.9314375921079544,
      "grad_norm": 0.27781397104263306,
      "hard_loss": 0.8389895732363039,
      "joint_kd_loss": 4.94124719383195,
      "learning_rate": 1.233383838208757e-07,
      "loss": 2.70972900390625,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 0.9637254155992099,
      "sre_geom_loss": 0.001909351324390638,
      "sre_kd_loss": 8.36319608022226,
      "sre_logit_loss": 8.204599075470615,
      "sre_span_loss": 0.06312938981689513,
      "step": 11100,
      "vision_kd_loss": 1.0374356380105019,
      "vision_logit_loss": 3.493514461070299
    },
    {
      "affinity_loss": 0.006550269323342945,
      "emkd_kd_loss": 1.563900014217943,
      "epoch": 0.9356332569372696,
      "grad_norm": 0.34851837158203125,
      "hard_loss": 0.8584683315865368,
      "joint_kd_loss": 4.661209230050445,
      "learning_rate": 1.0879164029799439e-07,
      "loss": 2.691962890625,
      "matched_vision_samples": 0.956875,
      "response_kd_loss": 1.034817163459656,
      "sre_geom_loss": 0.0020213122170832775,
      "sre_kd_loss": 7.75851844965131,
      "sre_logit_loss": 7.594015942535625,
      "sre_span_loss": 0.06343689980567432,
      "step": 11150,
      "vision_kd_loss": 1.046491433158517,
      "vision_logit_loss": 3.5309388008713722
    },
    {
      "affinity_loss": 0.006461084636248416,
      "emkd_kd_loss": 1.5352064257487654,
      "epoch": 0.9398289217665847,
      "grad_norm": 0.15524592995643616,
      "hard_loss": 0.8890608578068714,
      "joint_kd_loss": 5.52803272832185,
      "learning_rate": 9.514799204420332e-08,
      "loss": 2.909554443359375,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.0200169692457917,
      "sre_geom_loss": 0.001928086968079814,
      "sre_kd_loss": 9.520859001148493,
      "sre_logit_loss": 9.360361089551006,
      "sre_span_loss": 0.0640935684996657,
      "step": 11200,
      "vision_kd_loss": 1.0251979430392384,
      "vision_logit_loss": 3.454683314114809
    },
    {
      "affinity_loss": 0.006564175353414612,
      "emkd_kd_loss": 1.5725901100574993,
      "epoch": 0.9440245865958998,
      "grad_norm": 0.1180800348520279,
      "hard_loss": 0.8923234357354249,
      "joint_kd_loss": 3.776184835061431,
      "learning_rate": 8.240995817406284e-08,
      "loss": 2.47064697265625,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 1.0675793552348773,
      "sre_geom_loss": 0.001954924899724375,
      "sre_kd_loss": 5.97977955943672,
      "sre_logit_loss": 5.818433182857552,
      "sre_span_loss": 0.06360009735450149,
      "step": 11250,
      "vision_kd_loss": 1.0388004306703806,
      "vision_logit_loss": 3.498784184753895
    },
    {
      "affinity_loss": 0.0064987628470407795,
      "emkd_kd_loss": 1.5506603932194412,
      "epoch": 0.9482202514252149,
      "grad_norm": 0.09761125594377518,
      "hard_loss": 0.8644765591607984,
      "joint_kd_loss": 3.987453670985997,
      "learning_rate": 7.057989059272074e-08,
      "loss": 2.3286940002441407,
      "matched_vision_samples": 0.96,
      "response_kd_loss": 1.035412581411615,
      "sre_geom_loss": 0.0021154997399020913,
      "sre_kd_loss": 6.424246937902645,
      "sre_logit_loss": 6.255034727521124,
      "sre_span_loss": 0.06343723222846165,
      "step": 11300,
      "vision_kd_loss": 1.0329541051760316,
      "vision_logit_loss": 3.4819401386380195
    },
    {
      "affinity_loss": 0.006601653852849268,
      "emkd_kd_loss": 1.5644258689694106,
      "epoch": 0.95241591625453,
      "grad_norm": 0.23945939540863037,
      "hard_loss": 0.8725707789026228,
      "joint_kd_loss": 4.028485377170146,
      "learning_rate": 5.96599735616682e-08,
      "loss": 2.6132601928710937,
      "matched_vision_samples": 0.9525,
      "response_kd_loss": 1.0326816882294247,
      "sre_geom_loss": 0.0019814791711944223,
      "sre_kd_loss": 6.492544883766677,
      "sre_logit_loss": 6.329531205013772,
      "sre_span_loss": 0.0639397259359248,
      "step": 11350,
      "vision_kd_loss": 1.0480850231274963,
      "vision_logit_loss": 3.5321747028827666
    },
    {
      "affinity_loss": 0.006537676238658605,
      "emkd_kd_loss": 1.5515600904077291,
      "epoch": 0.9566115810838451,
      "grad_norm": 0.3371392488479614,
      "hard_loss": 0.847762668907626,
      "joint_kd_loss": 3.6781058203056456,
      "learning_rate": 4.9652223295445504e-08,
      "loss": 2.62465087890625,
      "matched_vision_samples": 0.95375,
      "response_kd_loss": 1.0328603162759418,
      "sre_geom_loss": 0.0019057293368983609,
      "sre_kd_loss": 5.80465156262042,
      "sre_logit_loss": 5.645961790367728,
      "sre_span_loss": 0.06340331176179462,
      "step": 11400,
      "vision_kd_loss": 1.035129932463169,
      "vision_logit_loss": 3.4867521065473555
    },
    {
      "affinity_loss": 0.006538372157956473,
      "emkd_kd_loss": 1.566883229566738,
      "epoch": 0.9608072459131602,
      "grad_norm": 0.07242666184902191,
      "hard_loss": 0.8458554139168973,
      "joint_kd_loss": 4.392118308190256,
      "learning_rate": 4.0558487589372756e-08,
      "loss": 2.4265943908691407,
      "matched_vision_samples": 0.950625,
      "response_kd_loss": 1.0533372907986278,
      "sre_geom_loss": 0.0020517669197866796,
      "sre_kd_loss": 7.217353400408756,
      "sre_logit_loss": 7.052189455978368,
      "sre_span_loss": 0.06257560329861007,
      "step": 11450,
      "vision_kd_loss": 1.0402145844325423,
      "vision_logit_loss": 3.507021123021841
    },
    {
      "affinity_loss": 0.006613149933546083,
      "emkd_kd_loss": 1.5633306909911335,
      "epoch": 0.9650029107424754,
      "grad_norm": 0.0971616581082344,
      "hard_loss": 0.852058654453831,
      "joint_kd_loss": 4.28792687729001,
      "learning_rate": 3.2380445478381037e-08,
      "loss": 2.5102114868164063,
      "matched_vision_samples": 0.95875,
      "response_kd_loss": 1.039685753604714,
      "sre_geom_loss": 0.0018548219998569238,
      "sre_kd_loss": 7.012523052762263,
      "sre_logit_loss": 6.856276537084923,
      "sre_span_loss": 0.06350538962869905,
      "step": 11500,
      "vision_kd_loss": 1.043487814925611,
      "vision_logit_loss": 3.512636264115572
    },
    {
      "affinity_loss": 0.006498355345538585,
      "emkd_kd_loss": 1.5273271402250976,
      "epoch": 0.9691985755717905,
      "grad_norm": 0.10716655850410461,
      "hard_loss": 0.8968514029657672,
      "joint_kd_loss": 3.619020199868828,
      "learning_rate": 2.511960692700377e-08,
      "loss": 2.3159368896484374,
      "matched_vision_samples": 0.960625,
      "response_kd_loss": 0.9796725266127595,
      "sre_geom_loss": 0.0018801054092891078,
      "sre_kd_loss": 5.7107133204839196,
      "sre_logit_loss": 5.554884435216373,
      "sre_span_loss": 0.06182354045216926,
      "step": 11550,
      "vision_kd_loss": 1.0374908770248294,
      "vision_logit_loss": 3.500127974152565
    },
    {
      "affinity_loss": 0.006508174761693226,
      "emkd_kd_loss": 1.5068715962581336,
      "epoch": 0.9733942404011056,
      "grad_norm": 0.2382117211818695,
      "hard_loss": 0.841440821910819,
      "joint_kd_loss": 3.935065433382988,
      "learning_rate": 1.877731255057691e-08,
      "loss": 2.8151409912109373,
      "matched_vision_samples": 0.955,
      "response_kd_loss": 0.9475737673370168,
      "sre_geom_loss": 0.0019412068095569167,
      "sre_kd_loss": 6.363259294395102,
      "sre_logit_loss": 6.20326291569072,
      "sre_span_loss": 0.0629359921393916,
      "step": 11600,
      "vision_kd_loss": 1.0330847118422388,
      "vision_logit_loss": 3.4815213705599306
    },
    {
      "affinity_loss": 0.006569312170322519,
      "emkd_kd_loss": 1.5545484135486185,
      "epoch": 0.9775899052304207,
      "grad_norm": 0.16128003597259521,
      "hard_loss": 0.8978044851341178,
      "joint_kd_loss": 3.4686985837109385,
      "learning_rate": 1.3354733367717532e-08,
      "loss": 2.4185931396484377,
      "matched_vision_samples": 0.954375,
      "response_kd_loss": 1.0237065505972203,
      "sre_geom_loss": 0.0020663285116486385,
      "sre_kd_loss": 5.382848785594105,
      "sre_logit_loss": 5.215896143261489,
      "sre_span_loss": 0.06363620106480085,
      "step": 11650,
      "vision_kd_loss": 1.0426951378583909,
      "vision_logit_loss": 3.5138493344187736
    },
    {
      "affinity_loss": 0.006447138533258112,
      "emkd_kd_loss": 1.5572899248544128,
      "epoch": 0.9817855700597358,
      "grad_norm": 0.10633111000061035,
      "hard_loss": 0.8840657689814043,
      "joint_kd_loss": 4.659627698231488,
      "learning_rate": 8.852870584108441e-09,
      "loss": 2.381152801513672,
      "matched_vision_samples": 0.941875,
      "response_kd_loss": 1.0625316058211438,
      "sre_geom_loss": 0.0019310437558572957,
      "sre_kd_loss": 7.761965435303282,
      "sre_logit_loss": 7.601595730714107,
      "sre_span_loss": 0.06381753652705811,
      "step": 11700,
      "vision_kd_loss": 1.0260241197422146,
      "vision_logit_loss": 3.45938262462616
    },
    {
      "affinity_loss": 0.006650069612514926,
      "emkd_kd_loss": 1.5313385897129774,
      "epoch": 0.9859812348890509,
      "grad_norm": 0.22130177915096283,
      "hard_loss": 0.8545012717865756,
      "joint_kd_loss": 4.286086948532611,
      "learning_rate": 5.2725554076404895e-09,
      "loss": 2.835530090332031,
      "matched_vision_samples": 0.964375,
      "response_kd_loss": 0.9626334142727905,
      "sre_geom_loss": 0.0018523589157935483,
      "sre_kd_loss": 7.040835301664192,
      "sre_logit_loss": 6.885591125531064,
      "sre_span_loss": 0.0626261984638404,
      "step": 11750,
      "vision_kd_loss": 1.0500218835100532,
      "vision_logit_loss": 3.5350805701315404
    },
    {
      "affinity_loss": 0.006514077340252698,
      "emkd_kd_loss": 1.5466038740891963,
      "epoch": 0.990176899718366,
      "grad_norm": 0.29645729064941406,
      "hard_loss": 0.8421182928374946,
      "joint_kd_loss": 4.45807266259566,
      "learning_rate": 2.6144488949392253e-09,
      "loss": 2.5624319458007814,
      "matched_vision_samples": 0.949375,
      "response_kd_loss": 1.012652468487421,
      "sre_geom_loss": 0.0019294580810549889,
      "sre_kd_loss": 7.369541457730811,
      "sre_logit_loss": 7.210074419441953,
      "sre_span_loss": 0.06299411698710174,
      "step": 11800,
      "vision_kd_loss": 1.040277641043067,
      "vision_logit_loss": 3.5097028310596943
    },
    {
      "affinity_loss": 0.00670281262442586,
      "emkd_kd_loss": 1.5811996431648732,
      "epoch": 0.9943725645476811,
      "grad_norm": 0.1795351654291153,
      "hard_loss": 0.8154480634225274,
      "joint_kd_loss": 2.7372695695981384,
      "learning_rate": 8.790418293136427e-10,
      "loss": 2.2469218444824217,
      "matched_vision_samples": 0.95875,
      "response_kd_loss": 1.0394621901203407,
      "sre_geom_loss": 0.0020354206064267456,
      "sre_kd_loss": 3.893339518944267,
      "sre_logit_loss": 3.7282970799655595,
      "sre_span_loss": 0.06327142033493147,
      "step": 11850,
      "vision_kd_loss": 1.061468550413847,
      "vision_logit_loss": 3.575592939853668
    },
    {
      "affinity_loss": 0.006466785418888321,
      "emkd_kd_loss": 1.5163215739419684,
      "epoch": 0.9985682293769962,
      "grad_norm": 0.11996354162693024,
      "hard_loss": 0.8803233522260143,
      "joint_kd_loss": 4.478534475993365,
      "learning_rate": 6.665463013422191e-11,
      "loss": 2.347486572265625,
      "matched_vision_samples": 0.944375,
      "response_kd_loss": 0.9816168662486814,
      "sre_geom_loss": 0.0020204890983138313,
      "sre_kd_loss": 7.440747379336972,
      "sre_logit_loss": 7.275772421984511,
      "sre_span_loss": 0.06395049076760188,
      "step": 11900,
      "vision_kd_loss": 1.0255131408944727,
      "vision_logit_loss": 3.455374022424221
    },
    {
      "affinity_loss": 0.006321482973568987,
      "emkd_kd_loss": 1.5171096175363212,
      "epoch": 1.0,
      "hard_loss": 0.8687137741020466,
      "joint_kd_loss": 3.9764671671739866,
      "matched_vision_samples": 0.9615384615384616,
      "response_kd_loss": 0.9863242236420747,
      "sre_geom_loss": 0.001733809200466058,
      "sre_kd_loss": 6.435824774649172,
      "sre_logit_loss": 6.288095184729772,
      "sre_span_loss": 0.06103924664945065,
      "step": 11918,
      "total_flos": 0.0,
      "train_loss": 1.488922280460261,
      "train_runtime": 169840.5442,
      "train_samples_per_second": 8.981,
      "train_steps_per_second": 0.07,
      "vision_kd_loss": 1.0239475039772061,
      "vision_logit_loss": 3.463641719067053
    }
  ],
  "logging_steps": 50,
  "max_steps": 11918,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}