{ "iter": 2000, "per_layer": { "time_mlp.1.weight": { "w_norm": 18.912269592285156, "g_norm": 0.12332244962453842, "shape": [ 1024, 256 ] }, "time_mlp.1.bias": { "w_norm": 1.1593341827392578, "g_norm": 0.010901880450546741, "shape": [ 1024 ] }, "time_mlp.3.weight": { "w_norm": 9.703208923339844, "g_norm": 0.01711881533265114, "shape": [ 256, 1024 ] }, "time_mlp.3.bias": { "w_norm": 0.2899875342845917, "g_norm": 0.006888911128044128, "shape": [ 256 ] }, "input_conv.weight": { "w_norm": 7.9453887939453125, "g_norm": 0.06517047435045242, "shape": [ 192, 6, 3, 3 ] }, "input_conv.bias": { "w_norm": 1.189807653427124, "g_norm": 0.02409413456916809, "shape": [ 192 ] }, "enc_blocks.0.norm1.weight": { "w_norm": 13.80312442779541, "g_norm": 0.007493571378290653, "shape": [ 192 ] }, "enc_blocks.0.norm1.bias": { "w_norm": 0.09649404138326645, "g_norm": 0.011275016702711582, "shape": [ 192 ] }, "enc_blocks.0.conv1.weight": { "w_norm": 8.424869537353516, "g_norm": 0.1024579405784607, "shape": [ 192, 192, 3, 3 ] }, "enc_blocks.0.conv1.bias": { "w_norm": 0.20213939249515533, "g_norm": 0.008418641053140163, "shape": [ 192 ] }, "enc_blocks.0.t_proj.weight": { "w_norm": 8.089444160461426, "g_norm": 0.012320656329393387, "shape": [ 192, 256 ] }, "enc_blocks.0.t_proj.bias": { "w_norm": 0.5025810599327087, "g_norm": 0.008417835459113121, "shape": [ 192 ] }, "enc_blocks.0.norm2.weight": { "w_norm": 13.835018157958984, "g_norm": 0.005644944962114096, "shape": [ 192 ] }, "enc_blocks.0.norm2.bias": { "w_norm": 0.06193859502673149, "g_norm": 0.0049712806940078735, "shape": [ 192 ] }, "enc_blocks.0.conv2.weight": { "w_norm": 8.313619613647461, "g_norm": 0.04298524931073189, "shape": [ 192, 192, 3, 3 ] }, "enc_blocks.0.conv2.bias": { "w_norm": 0.18812546133995056, "g_norm": 0.0026280551683157682, "shape": [ 192 ] }, "enc_blocks.1.norm1.weight": { "w_norm": 13.851577758789062, "g_norm": 6.118719466030598e-05, "shape": [ 192 ] }, "enc_blocks.1.norm1.bias": { "w_norm": 0.032818518579006195, "g_norm": 5.334233719622716e-05, "shape": [ 192 ] }, "enc_blocks.1.conv1.weight": { "w_norm": 8.230306625366211, "g_norm": 0.0018623622599989176, "shape": [ 192, 192, 3, 3 ] }, "enc_blocks.1.conv1.bias": { "w_norm": 0.19205407798290253, "g_norm": 8.16969622974284e-05, "shape": [ 192 ] }, "enc_blocks.1.t_proj.weight": { "w_norm": 8.11756420135498, "g_norm": 8.078471728367731e-05, "shape": [ 192, 256 ] }, "enc_blocks.1.t_proj.bias": { "w_norm": 0.5217092633247375, "g_norm": 8.170954242814332e-05, "shape": [ 192 ] }, "enc_blocks.1.norm2.weight": { "w_norm": 13.818221092224121, "g_norm": 0.00012724759289994836, "shape": [ 192 ] }, "enc_blocks.1.norm2.bias": { "w_norm": 0.05621594935655594, "g_norm": 7.47037265682593e-05, "shape": [ 192 ] }, "enc_blocks.1.conv2.weight": { "w_norm": 8.19234848022461, "g_norm": 0.002554898150265217, "shape": [ 192, 192, 3, 3 ] }, "enc_blocks.1.conv2.bias": { "w_norm": 0.208110049366951, "g_norm": 8.532928768545389e-05, "shape": [ 192 ] }, "enc_blocks.2.norm1.weight": { "w_norm": 13.827280044555664, "g_norm": 4.5593504182761535e-05, "shape": [ 192 ] }, "enc_blocks.2.norm1.bias": { "w_norm": 0.051892783492803574, "g_norm": 4.180220275884494e-05, "shape": [ 192 ] }, "enc_blocks.2.conv1.weight": { "w_norm": 11.666332244873047, "g_norm": 0.001202629879117012, "shape": [ 384, 192, 3, 3 ] }, "enc_blocks.2.conv1.bias": { "w_norm": 0.2665994167327881, "g_norm": 5.171901284484193e-05, "shape": [ 384 ] }, "enc_blocks.2.t_proj.weight": { "w_norm": 11.719046592712402, "g_norm": 4.928291673422791e-05, "shape": [ 384, 256 ] }, "enc_blocks.2.t_proj.bias": { "w_norm": 0.7221358418464661, "g_norm": 5.172061355551705e-05, "shape": [ 384 ] }, "enc_blocks.2.norm2.weight": { "w_norm": 19.450790405273438, "g_norm": 0.00012247661652509123, "shape": [ 384 ] }, "enc_blocks.2.norm2.bias": { "w_norm": 0.16962189972400665, "g_norm": 5.319947740645148e-05, "shape": [ 384 ] }, "enc_blocks.2.conv2.weight": { "w_norm": 11.663330078125, "g_norm": 0.001001948956400156, "shape": [ 384, 384, 3, 3 ] }, "enc_blocks.2.conv2.bias": { "w_norm": 0.20253244042396545, "g_norm": 2.679444514797069e-05, "shape": [ 384 ] }, "enc_blocks.2.skip.weight": { "w_norm": 11.206422805786133, "g_norm": 0.0009856950491666794, "shape": [ 384, 192, 1, 1 ] }, "enc_blocks.2.skip.bias": { "w_norm": 0.827253520488739, "g_norm": 2.679444514797069e-05, "shape": [ 384 ] }, "enc_blocks.3.norm1.weight": { "w_norm": 19.589677810668945, "g_norm": 2.0965567273378838e-06, "shape": [ 384 ] }, "enc_blocks.3.norm1.bias": { "w_norm": 0.04693835228681564, "g_norm": 2.1864386781089706e-06, "shape": [ 384 ] }, "enc_blocks.3.conv1.weight": { "w_norm": 11.792016983032227, "g_norm": 6.856802065158263e-05, "shape": [ 384, 384, 3, 3 ] }, "enc_blocks.3.conv1.bias": { "w_norm": 0.1976812332868576, "g_norm": 2.096104935844778e-06, "shape": [ 384 ] }, "enc_blocks.3.t_proj.weight": { "w_norm": 11.38448429107666, "g_norm": 2.964314489872777e-06, "shape": [ 384, 256 ] }, "enc_blocks.3.t_proj.bias": { "w_norm": 0.6924961805343628, "g_norm": 2.096133584927884e-06, "shape": [ 384 ] }, "enc_blocks.3.norm2.weight": { "w_norm": 19.44411277770996, "g_norm": 2.8147405828349292e-06, "shape": [ 384 ] }, "enc_blocks.3.norm2.bias": { "w_norm": 0.1529274880886078, "g_norm": 2.841748937498778e-06, "shape": [ 384 ] }, "enc_blocks.3.conv2.weight": { "w_norm": 11.47637939453125, "g_norm": 0.00019026761583518237, "shape": [ 384, 384, 3, 3 ] }, "enc_blocks.3.conv2.bias": { "w_norm": 0.1905771642923355, "g_norm": 6.41557244307478e-06, "shape": [ 384 ] }, "enc_blocks.4.norm1.weight": { "w_norm": 19.593414306640625, "g_norm": 1.7410387727068155e-06, "shape": [ 384 ] }, "enc_blocks.4.norm1.bias": { "w_norm": 0.05097125843167305, "g_norm": 1.8455435792930075e-06, "shape": [ 384 ] }, "enc_blocks.4.conv1.weight": { "w_norm": 16.576250076293945, "g_norm": 5.6877506722230464e-05, "shape": [ 768, 384, 3, 3 ] }, "enc_blocks.4.conv1.bias": { "w_norm": 0.28481438755989075, "g_norm": 1.893393459795334e-06, "shape": [ 768 ] }, "enc_blocks.4.t_proj.weight": { "w_norm": 16.05063247680664, "g_norm": 2.453651859468664e-06, "shape": [ 768, 256 ] }, "enc_blocks.4.t_proj.bias": { "w_norm": 0.9925292730331421, "g_norm": 1.8932430521090282e-06, "shape": [ 768 ] }, "enc_blocks.4.norm2.weight": { "w_norm": 27.487300872802734, "g_norm": 2.2128754153527552e-06, "shape": [ 768 ] }, "enc_blocks.4.norm2.bias": { "w_norm": 0.23648281395435333, "g_norm": 2.872084451155388e-06, "shape": [ 768 ] }, "enc_blocks.4.conv2.weight": { "w_norm": 16.297828674316406, "g_norm": 0.00014143867883831263, "shape": [ 768, 768, 3, 3 ] }, "enc_blocks.4.conv2.bias": { "w_norm": 0.19452296197414398, "g_norm": 4.4945381887373514e-06, "shape": [ 768 ] }, "enc_blocks.4.skip.weight": { "w_norm": 15.839735984802246, "g_norm": 0.00015103576879482716, "shape": [ 768, 384, 1, 1 ] }, "enc_blocks.4.skip.bias": { "w_norm": 0.8077318072319031, "g_norm": 4.4945381887373514e-06, "shape": [ 768 ] }, "enc_blocks.5.norm1.weight": { "w_norm": 27.712413787841797, "g_norm": 2.0804559426323976e-06, "shape": [ 768 ] }, "enc_blocks.5.norm1.bias": { "w_norm": 0.05509699136018753, "g_norm": 1.9732342479983345e-06, "shape": [ 768 ] }, "enc_blocks.5.conv1.weight": { "w_norm": 16.845355987548828, "g_norm": 4.273263766663149e-05, "shape": [ 768, 768, 3, 3 ] }, "enc_blocks.5.conv1.bias": { "w_norm": 0.2029583752155304, "g_norm": 8.974413958640071e-07, "shape": [ 768 ] }, "enc_blocks.5.t_proj.weight": { "w_norm": 16.008506774902344, "g_norm": 8.729576848054421e-07, "shape": [ 768, 256 ] }, "enc_blocks.5.t_proj.bias": { "w_norm": 0.9988090395927429, "g_norm": 8.977645507002308e-07, "shape": [ 768 ] }, "enc_blocks.5.norm2.weight": { "w_norm": 27.54796600341797, "g_norm": 2.552696969360113e-06, "shape": [ 768 ] }, "enc_blocks.5.norm2.bias": { "w_norm": 0.1847270131111145, "g_norm": 2.101767222484341e-06, "shape": [ 768 ] }, "enc_blocks.5.conv2.weight": { "w_norm": 16.440540313720703, "g_norm": 0.00010133799514733255, "shape": [ 768, 768, 3, 3 ] }, "enc_blocks.5.conv2.bias": { "w_norm": 0.19306622445583344, "g_norm": 2.4288849544973345e-06, "shape": [ 768 ] }, "enc_attns.3.norm.weight": { "w_norm": 19.538881301879883, "g_norm": 1.6037432715165778e-06, "shape": [ 384 ] }, "enc_attns.3.norm.bias": { "w_norm": 0.04634387046098709, "g_norm": 2.5036943043232895e-06, "shape": [ 384 ] }, "enc_attns.3.qkv.weight": { "w_norm": 19.671646118164062, "g_norm": 4.9293481424683705e-05, "shape": [ 1152, 384, 1, 1 ] }, "enc_attns.3.qkv.bias": { "w_norm": 1.0068951845169067, "g_norm": 3.873734385706484e-06, "shape": [ 1152 ] }, "enc_attns.3.proj.weight": { "w_norm": 11.309000968933105, "g_norm": 4.3482283217599615e-05, "shape": [ 384, 384, 1, 1 ] }, "enc_attns.3.proj.bias": { "w_norm": 0.5629428625106812, "g_norm": 6.388872861862183e-06, "shape": [ 384 ] }, "enc_attns.4.norm.weight": { "w_norm": 27.623943328857422, "g_norm": 1.3003150343138259e-06, "shape": [ 768 ] }, "enc_attns.4.norm.bias": { "w_norm": 0.048902470618486404, "g_norm": 1.8792874243445112e-06, "shape": [ 768 ] }, "enc_attns.4.qkv.weight": { "w_norm": 27.94339942932129, "g_norm": 5.037298615206964e-05, "shape": [ 2304, 768, 1, 1 ] }, "enc_attns.4.qkv.bias": { "w_norm": 1.0119142532348633, "g_norm": 2.70444297711947e-06, "shape": [ 2304 ] }, "enc_attns.4.proj.weight": { "w_norm": 16.00703239440918, "g_norm": 5.209520168136805e-05, "shape": [ 768, 768, 1, 1 ] }, "enc_attns.4.proj.bias": { "w_norm": 0.5728441476821899, "g_norm": 4.516856279224157e-06, "shape": [ 768 ] }, "enc_attns.5.norm.weight": { "w_norm": 27.662988662719727, "g_norm": 3.406920257020829e-07, "shape": [ 768 ] }, "enc_attns.5.norm.bias": { "w_norm": 0.04228585958480835, "g_norm": 8.539649343219935e-07, "shape": [ 768 ] }, "enc_attns.5.qkv.weight": { "w_norm": 27.92650604248047, "g_norm": 1.5280100342351943e-05, "shape": [ 2304, 768, 1, 1 ] }, "enc_attns.5.qkv.bias": { "w_norm": 0.9987878799438477, "g_norm": 1.4080259234106052e-06, "shape": [ 2304 ] }, "enc_attns.5.proj.weight": { "w_norm": 16.03470802307129, "g_norm": 1.587315637152642e-05, "shape": [ 768, 768, 1, 1 ] }, "enc_attns.5.proj.bias": { "w_norm": 0.5973420143127441, "g_norm": 2.459404413457378e-06, "shape": [ 768 ] }, "enc_downs.0.op.weight": { "w_norm": 8.154297828674316, "g_norm": 0.004857305437326431, "shape": [ 192, 192, 3, 3 ] }, "enc_downs.0.op.bias": { "w_norm": 0.18503881990909576, "g_norm": 9.78953976300545e-05, "shape": [ 192 ] }, "enc_downs.1.op.weight": { "w_norm": 8.09019947052002, "g_norm": 0.0018013118533417583, "shape": [ 192, 192, 3, 3 ] }, "enc_downs.1.op.bias": { "w_norm": 0.19495517015457153, "g_norm": 2.7523738026502542e-05, "shape": [ 192 ] }, "enc_downs.2.op.weight": { "w_norm": 11.496183395385742, "g_norm": 0.0004138555086683482, "shape": [ 384, 384, 3, 3 ] }, "enc_downs.2.op.bias": { "w_norm": 0.18931472301483154, "g_norm": 6.417965778382495e-06, "shape": [ 384 ] }, "enc_downs.3.op.weight": { "w_norm": 11.428959846496582, "g_norm": 0.00020830711582675576, "shape": [ 384, 384, 3, 3 ] }, "enc_downs.3.op.bias": { "w_norm": 0.1918579638004303, "g_norm": 2.7779085485235555e-06, "shape": [ 384 ] }, "enc_downs.4.op.weight": { "w_norm": 16.336414337158203, "g_norm": 0.0002319672203157097, "shape": [ 768, 768, 3, 3 ] }, "enc_downs.4.op.bias": { "w_norm": 0.1980322301387787, "g_norm": 2.4985531581478426e-06, "shape": [ 768 ] }, "mid1.norm1.weight": { "w_norm": 27.713109970092773, "g_norm": 5.832573606312508e-07, "shape": [ 768 ] }, "mid1.norm1.bias": { "w_norm": 0.04988333582878113, "g_norm": 4.770976715917641e-07, "shape": [ 768 ] }, "mid1.conv1.weight": { "w_norm": 16.79037094116211, "g_norm": 2.5980823920690455e-05, "shape": [ 768, 768, 3, 3 ] }, "mid1.conv1.bias": { "w_norm": 0.20654506981372833, "g_norm": 5.159295710654987e-07, "shape": [ 768 ] }, "mid1.t_proj.weight": { "w_norm": 16.01586151123047, "g_norm": 5.154277573637955e-07, "shape": [ 768, 256 ] }, "mid1.t_proj.bias": { "w_norm": 1.031219482421875, "g_norm": 5.159647002983547e-07, "shape": [ 768 ] }, "mid1.norm2.weight": { "w_norm": 27.551307678222656, "g_norm": 7.436812552441552e-07, "shape": [ 768 ] }, "mid1.norm2.bias": { "w_norm": 0.20324093103408813, "g_norm": 1.0805827059812145e-06, "shape": [ 768 ] }, "mid1.conv2.weight": { "w_norm": 16.460031509399414, "g_norm": 5.092180072097108e-05, "shape": [ 768, 768, 3, 3 ] }, "mid1.conv2.bias": { "w_norm": 0.1937713921070099, "g_norm": 1.829196662583854e-06, "shape": [ 768 ] }, "mid_attn.norm.weight": { "w_norm": 27.66978645324707, "g_norm": 2.9024468517491187e-07, "shape": [ 768 ] }, "mid_attn.norm.bias": { "w_norm": 0.044777318835258484, "g_norm": 6.936780891919625e-07, "shape": [ 768 ] }, "mid_attn.qkv.weight": { "w_norm": 27.8992919921875, "g_norm": 1.2187340871605556e-05, "shape": [ 2304, 768, 1, 1 ] }, "mid_attn.qkv.bias": { "w_norm": 0.9861307144165039, "g_norm": 1.0868640174521715e-06, "shape": [ 2304 ] }, "mid_attn.proj.weight": { "w_norm": 16.032732009887695, "g_norm": 1.2030658581352327e-05, "shape": [ 768, 768, 1, 1 ] }, "mid_attn.proj.bias": { "w_norm": 0.576869010925293, "g_norm": 1.8453860093359253e-06, "shape": [ 768 ] }, "mid2.norm1.weight": { "w_norm": 27.717756271362305, "g_norm": 1.05101287317666e-06, "shape": [ 768 ] }, "mid2.norm1.bias": { "w_norm": 0.06068818271160126, "g_norm": 1.0774595011753263e-06, "shape": [ 768 ] }, "mid2.conv1.weight": { "w_norm": 16.839242935180664, "g_norm": 2.6506331778364256e-05, "shape": [ 768, 768, 3, 3 ] }, "mid2.conv1.bias": { "w_norm": 0.19942057132720947, "g_norm": 5.999360723762948e-07, "shape": [ 768 ] }, "mid2.t_proj.weight": { "w_norm": 16.003231048583984, "g_norm": 6.345590009004809e-07, "shape": [ 768, 256 ] }, "mid2.t_proj.bias": { "w_norm": 0.9975320100784302, "g_norm": 5.998435312903894e-07, "shape": [ 768 ] }, "mid2.norm2.weight": { "w_norm": 27.539823532104492, "g_norm": 1.1429430060161394e-06, "shape": [ 768 ] }, "mid2.norm2.bias": { "w_norm": 0.2095635086297989, "g_norm": 1.4760953490622342e-06, "shape": [ 768 ] }, "mid2.conv2.weight": { "w_norm": 16.427698135375977, "g_norm": 4.7607252781745046e-05, "shape": [ 768, 768, 3, 3 ] }, "mid2.conv2.bias": { "w_norm": 0.19000190496444702, "g_norm": 1.873412543318409e-06, "shape": [ 768 ] }, "dec_blocks.0.norm1.weight": { "w_norm": 39.21333312988281, "g_norm": 1.0148718274649582e-06, "shape": [ 1536 ] }, "dec_blocks.0.norm1.bias": { "w_norm": 0.09431219100952148, "g_norm": 9.196940595757042e-07, "shape": [ 1536 ] }, "dec_blocks.0.conv1.weight": { "w_norm": 17.71900177001953, "g_norm": 3.547394589986652e-05, "shape": [ 768, 1536, 3, 3 ] }, "dec_blocks.0.conv1.bias": { "w_norm": 0.1529998481273651, "g_norm": 5.319739102560561e-07, "shape": [ 768 ] }, "dec_blocks.0.t_proj.weight": { "w_norm": 16.024818420410156, "g_norm": 6.486535539806937e-07, "shape": [ 768, 256 ] }, "dec_blocks.0.t_proj.bias": { "w_norm": 1.02411949634552, "g_norm": 5.319541855897114e-07, "shape": [ 768 ] }, "dec_blocks.0.norm2.weight": { "w_norm": 27.482181549072266, "g_norm": 1.8672670876185293e-06, "shape": [ 768 ] }, "dec_blocks.0.norm2.bias": { "w_norm": 0.2610810697078705, "g_norm": 3.00572719424963e-06, "shape": [ 768 ] }, "dec_blocks.0.conv2.weight": { "w_norm": 16.24523162841797, "g_norm": 9.572529961587861e-05, "shape": [ 768, 768, 3, 3 ] }, "dec_blocks.0.conv2.bias": { "w_norm": 0.19820284843444824, "g_norm": 4.452592747838935e-06, "shape": [ 768 ] }, "dec_blocks.0.skip.weight": { "w_norm": 15.814351081848145, "g_norm": 0.00045108215999789536, "shape": [ 768, 1536, 1, 1 ] }, "dec_blocks.0.skip.bias": { "w_norm": 0.4101272523403168, "g_norm": 4.452592747838935e-06, "shape": [ 768 ] }, "dec_blocks.1.norm1.weight": { "w_norm": 39.209964752197266, "g_norm": 3.3631847600190667e-06, "shape": [ 1536 ] }, "dec_blocks.1.norm1.bias": { "w_norm": 0.08060162514448166, "g_norm": 2.9157504286558833e-06, "shape": [ 1536 ] }, "dec_blocks.1.conv1.weight": { "w_norm": 17.587966918945312, "g_norm": 0.0001042117946781218, "shape": [ 768, 1536, 3, 3 ] }, "dec_blocks.1.conv1.bias": { "w_norm": 0.14630834758281708, "g_norm": 1.5931634607113665e-06, "shape": [ 768 ] }, "dec_blocks.1.t_proj.weight": { "w_norm": 16.054262161254883, "g_norm": 1.2944892660016194e-06, "shape": [ 768, 256 ] }, "dec_blocks.1.t_proj.bias": { "w_norm": 0.9880488514900208, "g_norm": 1.5929435903672129e-06, "shape": [ 768 ] }, "dec_blocks.1.norm2.weight": { "w_norm": 27.479366302490234, "g_norm": 3.6664212075265823e-06, "shape": [ 768 ] }, "dec_blocks.1.norm2.bias": { "w_norm": 0.2891330122947693, "g_norm": 5.97324697082513e-06, "shape": [ 768 ] }, "dec_blocks.1.conv2.weight": { "w_norm": 16.209659576416016, "g_norm": 0.00018095829000230879, "shape": [ 768, 768, 3, 3 ] }, "dec_blocks.1.conv2.bias": { "w_norm": 0.19671788811683655, "g_norm": 9.126384611590765e-06, "shape": [ 768 ] }, "dec_blocks.1.skip.weight": { "w_norm": 15.800586700439453, "g_norm": 0.0006704161060042679, "shape": [ 768, 1536, 1, 1 ] }, "dec_blocks.1.skip.bias": { "w_norm": 0.4059737026691437, "g_norm": 9.126384611590765e-06, "shape": [ 768 ] }, "dec_blocks.2.norm1.weight": { "w_norm": 33.95747375488281, "g_norm": 3.9997848944040015e-06, "shape": [ 1152 ] }, "dec_blocks.2.norm1.bias": { "w_norm": 0.08351629227399826, "g_norm": 3.245766492909752e-06, "shape": [ 1152 ] }, "dec_blocks.2.conv1.weight": { "w_norm": 12.456512451171875, "g_norm": 0.00016187748406082392, "shape": [ 384, 1152, 3, 3 ] }, "dec_blocks.2.conv1.bias": { "w_norm": 0.12158936262130737, "g_norm": 2.730666437855689e-06, "shape": [ 384 ] }, "dec_blocks.2.t_proj.weight": { "w_norm": 11.345858573913574, "g_norm": 2.7153444079885958e-06, "shape": [ 384, 256 ] }, "dec_blocks.2.t_proj.bias": { "w_norm": 0.7102914452552795, "g_norm": 2.730762389546726e-06, "shape": [ 384 ] }, "dec_blocks.2.norm2.weight": { "w_norm": 19.454023361206055, "g_norm": 7.419084795401432e-06, "shape": [ 384 ] }, "dec_blocks.2.norm2.bias": { "w_norm": 0.18189598619937897, "g_norm": 9.322840924141929e-06, "shape": [ 384 ] }, "dec_blocks.2.conv2.weight": { "w_norm": 11.435891151428223, "g_norm": 0.00031997592304833233, "shape": [ 384, 384, 3, 3 ] }, "dec_blocks.2.conv2.bias": { "w_norm": 0.18925972282886505, "g_norm": 1.760806844686158e-05, "shape": [ 384 ] }, "dec_blocks.2.skip.weight": { "w_norm": 11.040159225463867, "g_norm": 0.0008981034043245018, "shape": [ 384, 1152, 1, 1 ] }, "dec_blocks.2.skip.bias": { "w_norm": 0.3314064145088196, "g_norm": 1.760806844686158e-05, "shape": [ 384 ] }, "dec_blocks.3.norm1.weight": { "w_norm": 27.715042114257812, "g_norm": 2.8067692255717702e-05, "shape": [ 768 ] }, "dec_blocks.3.norm1.bias": { "w_norm": 0.07817665487527847, "g_norm": 2.4464015950798057e-05, "shape": [ 768 ] }, "dec_blocks.3.conv1.weight": { "w_norm": 12.271158218383789, "g_norm": 0.0007932300213724375, "shape": [ 384, 768, 3, 3 ] }, "dec_blocks.3.conv1.bias": { "w_norm": 0.14353975653648376, "g_norm": 1.6304338714689948e-05, "shape": [ 384 ] }, "dec_blocks.3.t_proj.weight": { "w_norm": 11.377923011779785, "g_norm": 1.7768579709809273e-05, "shape": [ 384, 256 ] }, "dec_blocks.3.t_proj.bias": { "w_norm": 0.7092573642730713, "g_norm": 1.6306374163832515e-05, "shape": [ 384 ] }, "dec_blocks.3.norm2.weight": { "w_norm": 19.434457778930664, "g_norm": 3.273187030572444e-05, "shape": [ 384 ] }, "dec_blocks.3.norm2.bias": { "w_norm": 0.18299466371536255, "g_norm": 3.3433821954531595e-05, "shape": [ 384 ] }, "dec_blocks.3.conv2.weight": { "w_norm": 11.477055549621582, "g_norm": 0.001047117984853685, "shape": [ 384, 384, 3, 3 ] }, "dec_blocks.3.conv2.bias": { "w_norm": 0.19348949193954468, "g_norm": 4.914805685984902e-05, "shape": [ 384 ] }, "dec_blocks.3.skip.weight": { "w_norm": 11.101613998413086, "g_norm": 0.0015457079280167818, "shape": [ 384, 768, 1, 1 ] }, "dec_blocks.3.skip.bias": { "w_norm": 0.40536269545555115, "g_norm": 4.914805685984902e-05, "shape": [ 384 ] }, "dec_blocks.4.norm1.weight": { "w_norm": 23.982229232788086, "g_norm": 7.216406083898619e-05, "shape": [ 576 ] }, "dec_blocks.4.norm1.bias": { "w_norm": 0.07750255614519119, "g_norm": 4.411262125358917e-05, "shape": [ 576 ] }, "dec_blocks.4.conv1.weight": { "w_norm": 8.488591194152832, "g_norm": 0.0018581050680950284, "shape": [ 192, 576, 3, 3 ] }, "dec_blocks.4.conv1.bias": { "w_norm": 0.11655846983194351, "g_norm": 4.4020147470291704e-05, "shape": [ 192 ] }, "dec_blocks.4.t_proj.weight": { "w_norm": 8.032074928283691, "g_norm": 4.8955851525533944e-05, "shape": [ 192, 256 ] }, "dec_blocks.4.t_proj.bias": { "w_norm": 0.48134493827819824, "g_norm": 4.4041335058864206e-05, "shape": [ 192 ] }, "dec_blocks.4.norm2.weight": { "w_norm": 13.797520637512207, "g_norm": 0.00010544883843977004, "shape": [ 192 ] }, "dec_blocks.4.norm2.bias": { "w_norm": 0.06829909980297089, "g_norm": 8.821025403449312e-05, "shape": [ 192 ] }, "dec_blocks.4.conv2.weight": { "w_norm": 8.082486152648926, "g_norm": 0.0036709108389914036, "shape": [ 192, 192, 3, 3 ] }, "dec_blocks.4.conv2.bias": { "w_norm": 0.19469857215881348, "g_norm": 0.00015675797476433218, "shape": [ 192 ] }, "dec_blocks.4.skip.weight": { "w_norm": 7.862843990325928, "g_norm": 0.006234226748347282, "shape": [ 192, 576, 1, 1 ] }, "dec_blocks.4.skip.bias": { "w_norm": 0.3469906747341156, "g_norm": 0.00015675797476433218, "shape": [ 192 ] }, "dec_blocks.5.norm1.weight": { "w_norm": 19.546398162841797, "g_norm": 0.0015075217233970761, "shape": [ 384 ] }, "dec_blocks.5.norm1.bias": { "w_norm": 0.10585547238588333, "g_norm": 0.001897464506328106, "shape": [ 384 ] }, "dec_blocks.5.conv1.weight": { "w_norm": 8.54923152923584, "g_norm": 0.02115660533308983, "shape": [ 192, 384, 3, 3 ] }, "dec_blocks.5.conv1.bias": { "w_norm": 0.13828550279140472, "g_norm": 0.000887448841240257, "shape": [ 192 ] }, "dec_blocks.5.t_proj.weight": { "w_norm": 8.180492401123047, "g_norm": 0.0015367768937721848, "shape": [ 192, 256 ] }, "dec_blocks.5.t_proj.bias": { "w_norm": 0.5051036477088928, "g_norm": 0.0008874782361090183, "shape": [ 192 ] }, "dec_blocks.5.norm2.weight": { "w_norm": 13.89153003692627, "g_norm": 0.001414273981936276, "shape": [ 192 ] }, "dec_blocks.5.norm2.bias": { "w_norm": 0.04068835452198982, "g_norm": 0.0013583751861006021, "shape": [ 192 ] }, "dec_blocks.5.conv2.weight": { "w_norm": 8.35645580291748, "g_norm": 0.018619170412421227, "shape": [ 192, 192, 3, 3 ] }, "dec_blocks.5.conv2.bias": { "w_norm": 0.20079921185970306, "g_norm": 0.0008557940018363297, "shape": [ 192 ] }, "dec_blocks.5.skip.weight": { "w_norm": 8.094463348388672, "g_norm": 0.024322308599948883, "shape": [ 192, 384, 1, 1 ] }, "dec_blocks.5.skip.bias": { "w_norm": 0.42241600155830383, "g_norm": 0.0008557940018363297, "shape": [ 192 ] }, "dec_attns.0.norm.weight": { "w_norm": 27.633703231811523, "g_norm": 2.9732186703768093e-06, "shape": [ 768 ] }, "dec_attns.0.norm.bias": { "w_norm": 0.03998072072863579, "g_norm": 3.1545473575533833e-06, "shape": [ 768 ] }, "dec_attns.0.qkv.weight": { "w_norm": 27.978946685791016, "g_norm": 8.574360981583595e-05, "shape": [ 2304, 768, 1, 1 ] }, "dec_attns.0.qkv.bias": { "w_norm": 1.0076439380645752, "g_norm": 3.600303898565471e-06, "shape": [ 2304 ] }, "dec_attns.0.proj.weight": { "w_norm": 16.07609748840332, "g_norm": 8.897064253687859e-05, "shape": [ 768, 768, 1, 1 ] }, "dec_attns.0.proj.bias": { "w_norm": 0.5759825706481934, "g_norm": 4.993305083189625e-06, "shape": [ 768 ] }, "dec_attns.1.norm.weight": { "w_norm": 27.627914428710938, "g_norm": 5.754376161348773e-06, "shape": [ 768 ] }, "dec_attns.1.norm.bias": { "w_norm": 0.04270288348197937, "g_norm": 6.391747774614487e-06, "shape": [ 768 ] }, "dec_attns.1.qkv.weight": { "w_norm": 27.995990753173828, "g_norm": 0.00017453204782214016, "shape": [ 2304, 768, 1, 1 ] }, "dec_attns.1.qkv.bias": { "w_norm": 1.0046143531799316, "g_norm": 7.337386250583222e-06, "shape": [ 2304 ] }, "dec_attns.1.proj.weight": { "w_norm": 16.055126190185547, "g_norm": 0.00015127500228118151, "shape": [ 768, 768, 1, 1 ] }, "dec_attns.1.proj.bias": { "w_norm": 0.5793938636779785, "g_norm": 1.0101544830831699e-05, "shape": [ 768 ] }, "dec_attns.2.norm.weight": { "w_norm": 19.51289939880371, "g_norm": 7.795668352628127e-06, "shape": [ 384 ] }, "dec_attns.2.norm.bias": { "w_norm": 0.025595588609576225, "g_norm": 1.0193104571953882e-05, "shape": [ 384 ] }, "dec_attns.2.qkv.weight": { "w_norm": 19.6967830657959, "g_norm": 0.00019536222680471838, "shape": [ 1152, 384, 1, 1 ] }, "dec_attns.2.qkv.bias": { "w_norm": 0.9850842356681824, "g_norm": 1.2773794878739864e-05, "shape": [ 1152 ] }, "dec_attns.2.proj.weight": { "w_norm": 11.304216384887695, "g_norm": 0.00019062748469877988, "shape": [ 384, 384, 1, 1 ] }, "dec_attns.2.proj.bias": { "w_norm": 0.5871286988258362, "g_norm": 1.9087274267803878e-05, "shape": [ 384 ] }, "dec_ups.0.op.weight": { "w_norm": 16.16544532775879, "g_norm": 0.000586340669542551, "shape": [ 768, 768, 3, 3 ] }, "dec_ups.0.op.bias": { "w_norm": 0.2004891335964203, "g_norm": 4.288574018573854e-06, "shape": [ 768 ] }, "dec_ups.1.op.weight": { "w_norm": 16.172229766845703, "g_norm": 0.0009024080936796963, "shape": [ 768, 768, 3, 3 ] }, "dec_ups.1.op.bias": { "w_norm": 0.19070285558700562, "g_norm": 8.297952263092156e-06, "shape": [ 768 ] }, "dec_ups.2.op.weight": { "w_norm": 11.33031177520752, "g_norm": 0.0016408424125984311, "shape": [ 384, 384, 3, 3 ] }, "dec_ups.2.op.bias": { "w_norm": 0.1964583396911621, "g_norm": 2.5792473024921492e-05, "shape": [ 384 ] }, "dec_ups.3.op.weight": { "w_norm": 11.363648414611816, "g_norm": 0.0027039572596549988, "shape": [ 384, 384, 3, 3 ] }, "dec_ups.3.op.bias": { "w_norm": 0.19799400866031647, "g_norm": 6.683699757559225e-05, "shape": [ 384 ] }, "dec_ups.4.op.weight": { "w_norm": 8.007488250732422, "g_norm": 0.011867591179907322, "shape": [ 192, 192, 3, 3 ] }, "dec_ups.4.op.bias": { "w_norm": 0.19662714004516602, "g_norm": 0.00022351545339915901, "shape": [ 192 ] }, "out_norm.weight": { "w_norm": 13.781282424926758, "g_norm": 0.005015091970562935, "shape": [ 192 ] }, "out_norm.bias": { "w_norm": 0.09246864169836044, "g_norm": 0.005548957269638777, "shape": [ 192 ] }, "out_conv.weight": { "w_norm": 0.1450980007648468, "g_norm": 0.3885398805141449, "shape": [ 3, 192, 3, 3 ] }, "out_conv.bias": { "w_norm": 0.008651298470795155, "g_norm": 0.023214390501379967, "shape": [ 3 ] } }, "activations": { "input_conv": { "mean": 0.007205421570688486, "std": 0.4401821494102478, "abs_max": 3.21875, "frac_dead": 0.0004186829028185457, "frac_sat": 0.0 }, "time_mlp.1": { "mean": 0.017198296263813972, "std": 0.2722778022289276, "abs_max": 3.90625, "frac_dead": 0.0, "frac_sat": 0.0 }, "time_mlp.3": { "mean": 0.025112560018897057, "std": 0.20933035016059875, "abs_max": 0.98828125, "frac_dead": 0.00032552084303461015, "frac_sat": 0.0 }, "enc_blocks.0.norm1": { "mean": -0.005679808557033539, "std": 0.9958928823471069, "abs_max": 6.680795669555664, "frac_dead": 6.159146437312302e-07, "frac_sat": 0.0 }, "enc_blocks.0.conv1": { "mean": 0.08074763417243958, "std": 0.9407960772514343, "abs_max": 10.6875, "frac_dead": 4.441208511707373e-05, "frac_sat": 2.6490953430879927e-08 }, "enc_blocks.0.t_proj": { "mean": -0.0006273089675232768, "std": 0.21317224204540253, "abs_max": 2.125, "frac_dead": 0.0, "frac_sat": 0.0 }, "enc_blocks.0.norm2": { "mean": -0.004103214479982853, "std": 0.9992560148239136, "abs_max": 9.351706504821777, "frac_dead": 8.675787057654816e-07, "frac_sat": 0.0 }, "enc_blocks.0.conv2": { "mean": -0.06078268960118294, "std": 0.9260948896408081, "abs_max": 11.4375, "frac_dead": 3.555748116923496e-05, "frac_sat": 8.278423280216884e-08 }, "enc_downs.0.op": { "mean": 0.02630145289003849, "std": 1.916042447090149, "abs_max": 22.75, "frac_dead": 2.0609961211448535e-05, "frac_sat": 0.0005958874826319516 }, "enc_blocks.1.norm1": { "mean": -0.0018810041947290301, "std": 1.0003385543823242, "abs_max": 10.622716903686523, "frac_dead": 1.0596381798677612e-06, "frac_sat": 2.6490953430879927e-08 }, "enc_blocks.1.conv1": { "mean": -0.0316561721265316, "std": 0.93098384141922, "abs_max": 11.8125, "frac_dead": 3.63720791938249e-05, "frac_sat": 2.649095449669403e-07 }, "enc_blocks.1.t_proj": { "mean": 0.004489074926823378, "std": 0.19143736362457275, "abs_max": 1.515625, "frac_dead": 0.0, "frac_sat": 0.0 }, "enc_blocks.1.norm2": { "mean": -0.004230222199112177, "std": 0.9972831010818481, "abs_max": 12.254587173461914, "frac_dead": 1.0066562481370056e-06, "frac_sat": 1.1920928955078125e-07 }, "enc_blocks.1.conv2": { "mean": -0.01853681169450283, "std": 0.78462815284729, "abs_max": 9.9375, "frac_dead": 4.1166942537529394e-05, "frac_sat": 0.0 }, "enc_downs.1.op": { "mean": -0.061043038964271545, "std": 2.2881019115448, "abs_max": 19.125, "frac_dead": 1.5523699403274804e-05, "frac_sat": 0.0006492932443507016 }, "enc_blocks.2.norm1": { "mean": -0.0027286489494144917, "std": 0.998390793800354, "abs_max": 8.235820770263672, "frac_dead": 6.887647714393097e-07, "frac_sat": 0.0 }, "enc_blocks.2.conv1": { "mean": -0.08958557993173599, "std": 1.2522926330566406, "abs_max": 18.125, "frac_dead": 3.594822555896826e-05, "frac_sat": 0.00014895862841513008 }, "enc_blocks.2.t_proj": { "mean": -0.004121434409171343, "std": 0.2775437831878662, "abs_max": 2.515625, "frac_dead": 0.0, "frac_sat": 0.0 }, "enc_blocks.2.norm2": { "mean": -0.009121330454945564, "std": 0.9945322275161743, "abs_max": 13.316609382629395, "frac_dead": 1.1656019296424347e-06, "frac_sat": 3.708733402163489e-06 }, "enc_blocks.2.conv2": { "mean": -0.0012582726776599884, "std": 0.8416869640350342, "abs_max": 8.625, "frac_dead": 2.6332007109886035e-05, "frac_sat": 0.0 }, "enc_blocks.2.skip": { "mean": 0.03872901201248169, "std": 1.1367685794830322, "abs_max": 14.8125, "frac_dead": 9.706285345600918e-05, "frac_sat": 7.70886708778562e-06 }, "enc_downs.2.op": { "mean": 0.11237926036119461, "std": 1.6770859956741333, "abs_max": 18.0, "frac_dead": 1.229180270456709e-05, "frac_sat": 8.911556506063789e-05 }, "enc_blocks.3.norm1": { "mean": -0.00042686983942985535, "std": 1.0004702806472778, "abs_max": 7.873823165893555, "frac_dead": 5.298190899338806e-07, "frac_sat": 0.0 }, "enc_blocks.3.conv1": { "mean": -0.19867272675037384, "std": 1.476852297782898, "abs_max": 18.0, "frac_dead": 2.076890814350918e-05, "frac_sat": 0.00023947822046466172 }, "enc_blocks.3.t_proj": { "mean": 0.002406445099040866, "std": 0.13080888986587524, "abs_max": 1.0390625, "frac_dead": 0.0, "frac_sat": 0.0 }, "enc_blocks.3.norm2": { "mean": -0.008266602642834187, "std": 0.9931374788284302, "abs_max": 11.833065032958984, "frac_dead": 1.271565793103946e-06, "frac_sat": 3.814697265625e-06 }, "enc_blocks.3.conv2": { "mean": 0.04491338133811951, "std": 0.7213199138641357, "abs_max": 6.40625, "frac_dead": 3.041161471628584e-05, "frac_sat": 0.0 }, "enc_attns.3.norm": { "mean": 0.0003345790901221335, "std": 0.997136116027832, "abs_max": 8.509593963623047, "frac_dead": 1.6954210195763153e-06, "frac_sat": 0.0 }, "enc_attns.3.qkv": { "mean": 0.0020568061154335737, "std": 0.606049656867981, "abs_max": 5.75, "frac_dead": 0.00010850694525288418, "frac_sat": 0.0 }, "enc_attns.3.proj": { "mean": -0.004439083393663168, "std": 0.24444590508937836, "abs_max": 1.0546875, "frac_dead": 0.0002475314831826836, "frac_sat": 0.0 }, "enc_downs.3.op": { "mean": -0.11371082812547684, "std": 2.028658151626587, "abs_max": 20.875, "frac_dead": 1.2715657248918433e-05, "frac_sat": 0.0005972120561636984 }, "enc_blocks.4.norm1": { "mean": 0.00025804672623053193, "std": 1.001221776008606, "abs_max": 7.8459954261779785, "frac_dead": 2.1192763597355224e-06, "frac_sat": 0.0 }, "enc_blocks.4.conv1": { "mean": -0.3934686481952667, "std": 1.384687900543213, "abs_max": 17.25, "frac_dead": 2.3100112230167724e-05, "frac_sat": 0.00015407138562295586 }, "enc_blocks.4.t_proj": { "mean": -0.002096842974424362, "std": 0.09800918400287628, "abs_max": 0.6796875, "frac_dead": 0.0, "frac_sat": 0.0 }, "enc_blocks.4.norm2": { "mean": -0.008577298372983932, "std": 0.9937466382980347, "abs_max": 10.411909103393555, "frac_dead": 2.119276274470394e-07, "frac_sat": 8.477105097881577e-07 }, "enc_blocks.4.conv2": { "mean": -0.012939635664224625, "std": 0.9054562449455261, "abs_max": 5.40625, "frac_dead": 1.5894571333774365e-05, "frac_sat": 0.0 }, "enc_blocks.4.skip": { "mean": -0.05520473048090935, "std": 1.2102307081222534, "abs_max": 12.3125, "frac_dead": 6.37902194284834e-05, "frac_sat": 6.781684078305261e-06 }, "enc_attns.4.norm": { "mean": -1.8503931642044336e-05, "std": 0.9971495866775513, "abs_max": 7.199902057647705, "frac_dead": 1.0596381798677612e-06, "frac_sat": 0.0 }, "enc_attns.4.qkv": { "mean": -0.004289580043405294, "std": 0.7320247292518616, "abs_max": 8.125, "frac_dead": 6.859390850877389e-05, "frac_sat": 0.0 }, "enc_attns.4.proj": { "mean": 0.002922792686149478, "std": 0.3093913197517395, "abs_max": 2.65625, "frac_dead": 0.0001275804388569668, "frac_sat": 0.0 }, "enc_downs.4.op": { "mean": 0.02217293158173561, "std": 3.0861616134643555, "abs_max": 25.875, "frac_dead": 2.543131586207892e-06, "frac_sat": 0.007915073074400425 }, "enc_blocks.5.norm1": { "mean": -8.908328163670376e-05, "std": 1.000154733657837, "abs_max": 7.034839153289795, "frac_dead": 1.6954210195763153e-06, "frac_sat": 0.0 }, "enc_blocks.5.conv1": { "mean": -0.2771743834018707, "std": 2.046868324279785, "abs_max": 19.25, "frac_dead": 1.3563368156610522e-05, "frac_sat": 0.0013817681465297937 }, "enc_blocks.5.t_proj": { "mean": -0.0009553485433571041, "std": 0.08868180215358734, "abs_max": 0.62890625, "frac_dead": 0.0, "frac_sat": 0.0 }, "enc_blocks.5.norm2": { "mean": -0.005851236637681723, "std": 0.9946741461753845, "abs_max": 8.874484062194824, "frac_dead": 8.477105097881577e-07, "frac_sat": 0.0 }, "enc_blocks.5.conv2": { "mean": 0.0862385705113411, "std": 1.4481452703475952, "abs_max": 8.5625, "frac_dead": 1.3563368156610522e-05, "frac_sat": 0.0 }, "enc_attns.5.norm": { "mean": 5.221697938395664e-05, "std": 0.9983348846435547, "abs_max": 7.151754379272461, "frac_dead": 2.543131586207892e-06, "frac_sat": 0.0 }, "enc_attns.5.qkv": { "mean": -0.004897332284599543, "std": 0.8810105323791504, "abs_max": 8.375, "frac_dead": 6.357828533509746e-05, "frac_sat": 0.0 }, "enc_attns.5.proj": { "mean": -0.006027805618941784, "std": 0.19361162185668945, "abs_max": 1.53125, "frac_dead": 0.0002526177268009633, "frac_sat": 0.0 }, "mid1.norm1": { "mean": 0.00011206004273844883, "std": 1.0003716945648193, "abs_max": 7.116409778594971, "frac_dead": 8.477105097881577e-07, "frac_sat": 0.0 }, "mid1.conv1": { "mean": -0.7086418867111206, "std": 1.8661444187164307, "abs_max": 21.0, "frac_dead": 1.52587890625e-05, "frac_sat": 0.003971523605287075 }, "mid1.t_proj": { "mean": 0.002929791109636426, "std": 0.08738984912633896, "abs_max": 0.5078125, "frac_dead": 5.425347262644209e-05, "frac_sat": 0.0 }, "mid1.norm2": { "mean": -0.007256262004375458, "std": 0.9950929880142212, "abs_max": 10.638596534729004, "frac_dead": 8.477105097881577e-07, "frac_sat": 1.6954210195763153e-06 }, "mid1.conv2": { "mean": 0.03888840973377228, "std": 0.7718620300292969, "abs_max": 4.5, "frac_dead": 2.204047268605791e-05, "frac_sat": 0.0 }, "mid_attn.norm": { "mean": 0.00013342582678887993, "std": 0.9982437491416931, "abs_max": 6.748208045959473, "frac_dead": 8.477105097881577e-07, "frac_sat": 0.0 }, "mid_attn.qkv": { "mean": -0.010101129300892353, "std": 0.871902585029602, "abs_max": 7.5, "frac_dead": 5.8774596254806966e-05, "frac_sat": 0.0 }, "mid_attn.proj": { "mean": -0.004268778953701258, "std": 0.17432516813278198, "abs_max": 1.2734375, "frac_dead": 0.00027974447584711015, "frac_sat": 0.0 }, "mid2.norm1": { "mean": 0.0004497667250689119, "std": 1.0008673667907715, "abs_max": 6.641637325286865, "frac_dead": 8.477105097881577e-07, "frac_sat": 0.0 }, "mid2.conv1": { "mean": -0.9663907289505005, "std": 2.2991340160369873, "abs_max": 24.25, "frac_dead": 2.373589450144209e-05, "frac_sat": 0.01361083984375 }, "mid2.t_proj": { "mean": 0.0009166261297650635, "std": 0.0890422984957695, "abs_max": 0.6640625, "frac_dead": 0.0, "frac_sat": 0.0 }, "mid2.norm2": { "mean": -0.00821609515696764, "std": 0.9965092539787292, "abs_max": 10.094039916992188, "frac_dead": 8.477105097881577e-07, "frac_sat": 8.477105097881577e-07 }, "mid2.conv2": { "mean": 0.08653509616851807, "std": 1.1487525701522827, "abs_max": 5.75, "frac_dead": 1.9497341781971045e-05, "frac_sat": 0.0 }, "dec_blocks.0.norm1": { "mean": 0.0010238753166049719, "std": 1.0017040967941284, "abs_max": 7.182895660400391, "frac_dead": 8.477105097881577e-07, "frac_sat": 0.0 }, "dec_blocks.0.conv1": { "mean": -2.5499513149261475, "std": 5.234790802001953, "abs_max": 55.25, "frac_dead": 5.086263172415784e-06, "frac_sat": 0.0761905238032341 }, "dec_blocks.0.t_proj": { "mean": -0.0014294469729065895, "std": 0.09479585289955139, "abs_max": 0.60546875, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_blocks.0.norm2": { "mean": -0.010944444686174393, "std": 0.9966337084770203, "abs_max": 10.068451881408691, "frac_dead": 2.543131586207892e-06, "frac_sat": 1.6954210195763153e-06 }, "dec_blocks.0.conv2": { "mean": 0.011901033110916615, "std": 0.5798095464706421, "abs_max": 3.125, "frac_dead": 2.7126736313221045e-05, "frac_sat": 0.0 }, "dec_blocks.0.skip": { "mean": -0.044557470828294754, "std": 1.870935082435608, "abs_max": 14.6875, "frac_dead": 1.610649997019209e-05, "frac_sat": 0.00020429823780432343 }, "dec_attns.0.norm": { "mean": -3.814531737589277e-05, "std": 0.9969810247421265, "abs_max": 6.740102767944336, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_attns.0.qkv": { "mean": -0.007594096474349499, "std": 0.7095833420753479, "abs_max": 4.78125, "frac_dead": 7.262053259182721e-05, "frac_sat": 0.0 }, "dec_attns.0.proj": { "mean": 0.02804269827902317, "std": 0.650712788105011, "abs_max": 2.953125, "frac_dead": 6.442599988076836e-05, "frac_sat": 0.0 }, "dec_ups.0.op": { "mean": 0.01264216285198927, "std": 2.6158359050750732, "abs_max": 23.0, "frac_dead": 6.357828624459216e-06, "frac_sat": 0.001959271030500531 }, "dec_blocks.1.norm1": { "mean": 0.0003692789177875966, "std": 1.0006011724472046, "abs_max": 7.659249305725098, "frac_dead": 5.298190899338806e-07, "frac_sat": 0.0 }, "dec_blocks.1.conv1": { "mean": -2.1808557510375977, "std": 3.315422534942627, "abs_max": 48.0, "frac_dead": 5.510118171514478e-06, "frac_sat": 0.03862402215600014 }, "dec_blocks.1.t_proj": { "mean": -0.0016897112363949418, "std": 0.09164174646139145, "abs_max": 0.6328125, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_blocks.1.norm2": { "mean": -0.012301405891776085, "std": 0.9972518086433411, "abs_max": 11.372942924499512, "frac_dead": 6.35782896551973e-07, "frac_sat": 7.841322258173022e-06 }, "dec_blocks.1.conv2": { "mean": -0.0036889174953103065, "std": 0.41822341084480286, "abs_max": 2.515625, "frac_dead": 3.708733493112959e-05, "frac_sat": 0.0 }, "dec_blocks.1.skip": { "mean": -0.013397175818681717, "std": 1.399378776550293, "abs_max": 10.625, "frac_dead": 2.7974447220913135e-05, "frac_sat": 6.35782896551973e-07 }, "dec_attns.1.norm": { "mean": -2.5563769668224268e-05, "std": 0.996963381767273, "abs_max": 6.62148904800415, "frac_dead": 6.35782896551973e-07, "frac_sat": 0.0 }, "dec_attns.1.qkv": { "mean": 0.0024721473455429077, "std": 0.6335548758506775, "abs_max": 4.03125, "frac_dead": 7.466916576959193e-05, "frac_sat": 0.0 }, "dec_attns.1.proj": { "mean": -0.008538656868040562, "std": 0.36578020453453064, "abs_max": 1.3671875, "frac_dead": 0.00012969970703125, "frac_sat": 0.0 }, "dec_ups.1.op": { "mean": 0.045211486518383026, "std": 1.5917174816131592, "abs_max": 12.5, "frac_dead": 9.324815437139478e-06, "frac_sat": 1.88085778063396e-05 }, "dec_blocks.2.norm1": { "mean": -3.3550793887116015e-05, "std": 1.0003987550735474, "abs_max": 8.471176147460938, "frac_dead": 8.477105097881577e-07, "frac_sat": 0.0 }, "dec_blocks.2.conv1": { "mean": -1.0419565439224243, "std": 3.1819097995758057, "abs_max": 34.75, "frac_dead": 9.960598617908545e-06, "frac_sat": 0.03538068011403084 }, "dec_blocks.2.t_proj": { "mean": -0.0044543445110321045, "std": 0.10459078103303909, "abs_max": 0.62109375, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_blocks.2.norm2": { "mean": -0.011841188184916973, "std": 1.0002442598342896, "abs_max": 9.476700782775879, "frac_dead": 2.4371677227463806e-06, "frac_sat": 0.0 }, "dec_blocks.2.conv2": { "mean": -0.003868026891723275, "std": 0.3635369539260864, "abs_max": 2.078125, "frac_dead": 5.647871148539707e-05, "frac_sat": 0.0 }, "dec_blocks.2.skip": { "mean": 0.0852733850479126, "std": 1.2003114223480225, "abs_max": 9.1875, "frac_dead": 3.178914266754873e-05, "frac_sat": 0.0 }, "dec_attns.2.norm": { "mean": 7.900595664978027e-05, "std": 0.9957061409950256, "abs_max": 6.682980060577393, "frac_dead": 2.119276274470394e-07, "frac_sat": 0.0 }, "dec_attns.2.qkv": { "mean": 0.013404250144958496, "std": 0.6031450629234314, "abs_max": 4.625, "frac_dead": 9.716882050270215e-05, "frac_sat": 0.0 }, "dec_attns.2.proj": { "mean": -0.02518179453909397, "std": 0.3698141872882843, "abs_max": 1.265625, "frac_dead": 0.00016360812878701836, "frac_sat": 0.0 }, "dec_ups.2.op": { "mean": 0.03491879999637604, "std": 1.0012180805206299, "abs_max": 7.09375, "frac_dead": 2.278222018503584e-05, "frac_sat": 0.0 }, "dec_blocks.3.norm1": { "mean": -0.0005132730002515018, "std": 1.0001797676086426, "abs_max": 10.262784957885742, "frac_dead": 7.682376690354431e-07, "frac_sat": 1.3245476715439963e-08 }, "dec_blocks.3.conv1": { "mean": -0.8485596776008606, "std": 2.129643440246582, "abs_max": 34.5, "frac_dead": 1.5709136278019287e-05, "frac_sat": 0.007686244323849678 }, "dec_blocks.3.t_proj": { "mean": -0.00039477564860135317, "std": 0.12362557649612427, "abs_max": 0.7734375, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_blocks.3.norm2": { "mean": -0.011190549470484257, "std": 0.995294988155365, "abs_max": 12.03353500366211, "frac_dead": 2.278221927554114e-06, "frac_sat": 1.141760094469646e-05 }, "dec_blocks.3.conv2": { "mean": -0.0035135315265506506, "std": 0.3895057141780853, "abs_max": 5.3125, "frac_dead": 5.745887756347656e-05, "frac_sat": 0.0 }, "dec_blocks.3.skip": { "mean": 0.03638695552945137, "std": 0.7669461369514465, "abs_max": 7.1875, "frac_dead": 5.66906419408042e-05, "frac_sat": 0.0 }, "dec_ups.3.op": { "mean": -0.04504766687750816, "std": 0.8403801321983337, "abs_max": 8.75, "frac_dead": 2.6305517167202197e-05, "frac_sat": 0.0 }, "dec_blocks.4.norm1": { "mean": -0.0013438367750495672, "std": 1.000179409980774, "abs_max": 11.959559440612793, "frac_dead": 9.757501402418711e-07, "frac_sat": 1.3686992872408155e-07 }, "dec_blocks.4.conv1": { "mean": -0.5712231397628784, "std": 1.8479676246643066, "abs_max": 22.875, "frac_dead": 1.3576614037447143e-05, "frac_sat": 0.0011172691592946649 }, "dec_blocks.4.t_proj": { "mean": -0.004008995834738016, "std": 0.12154782563447952, "abs_max": 0.6875, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_blocks.4.norm2": { "mean": -0.005222403444349766, "std": 0.996995210647583, "abs_max": 11.499641418457031, "frac_dead": 1.4437570143854828e-06, "frac_sat": 7.947286206899662e-08 }, "dec_blocks.4.conv2": { "mean": -0.008350415155291557, "std": 0.5789016485214233, "abs_max": 5.625, "frac_dead": 5.3591200412483886e-05, "frac_sat": 0.0 }, "dec_blocks.4.skip": { "mean": 0.009770325385034084, "std": 1.1018366813659668, "abs_max": 8.875, "frac_dead": 5.510118353413418e-05, "frac_sat": 0.0 }, "dec_ups.4.op": { "mean": -0.18501953780651093, "std": 2.0624237060546875, "abs_max": 22.75, "frac_dead": 2.249413046229165e-05, "frac_sat": 0.002946823835372925 }, "dec_blocks.5.norm1": { "mean": -0.003984882961958647, "std": 0.9979493021965027, "abs_max": 12.960043907165527, "frac_dead": 1.1656019296424347e-06, "frac_sat": 4.106097719613899e-07 }, "dec_blocks.5.conv1": { "mean": -0.0586412288248539, "std": 1.5150682926177979, "abs_max": 22.75, "frac_dead": 1.675883868301753e-05, "frac_sat": 0.0001257459371117875 }, "dec_blocks.5.t_proj": { "mean": 0.011308521963655949, "std": 0.215815931558609, "abs_max": 1.6015625, "frac_dead": 0.0, "frac_sat": 0.0 }, "dec_blocks.5.norm2": { "mean": 0.0006456120754592121, "std": 1.0051116943359375, "abs_max": 17.839805603027344, "frac_dead": 1.1026859283447266e-06, "frac_sat": 4.159079708188074e-06 }, "dec_blocks.5.conv2": { "mean": -0.9917439222335815, "std": 1.8552374839782715, "abs_max": 27.125, "frac_dead": 2.7272437364445068e-05, "frac_sat": 0.0017922355327755213 }, "dec_blocks.5.skip": { "mean": -0.26965466141700745, "std": 1.6131691932678223, "abs_max": 19.375, "frac_dead": 5.5223703384399414e-05, "frac_sat": 0.00018515189003665 }, "out_norm": { "mean": -0.004453430417925119, "std": 0.9956813454627991, "abs_max": 13.229631423950195, "frac_dead": 1.8510553445594269e-06, "frac_sat": 3.7418470810735016e-07 }, "out_conv": { "mean": 0.49139833450317383, "std": 0.23420317471027374, "abs_max": 1.046875, "frac_dead": 3.602769766075653e-06, "frac_sat": 0.0 } } }