| { |
| "iter": 2500, |
| "per_layer": { |
| "time_mlp.1.weight": { |
| "w_norm": 18.984146118164062, |
| "g_norm": 0.06113049387931824, |
| "shape": [ |
| 1024, |
| 256 |
| ] |
| }, |
| "time_mlp.1.bias": { |
| "w_norm": 1.159756064414978, |
| "g_norm": 0.005403383634984493, |
| "shape": [ |
| 1024 |
| ] |
| }, |
| "time_mlp.3.weight": { |
| "w_norm": 9.721464157104492, |
| "g_norm": 0.00959283672273159, |
| "shape": [ |
| 256, |
| 1024 |
| ] |
| }, |
| "time_mlp.3.bias": { |
| "w_norm": 0.2891911566257477, |
| "g_norm": 0.0048805121332407, |
| "shape": [ |
| 256 |
| ] |
| }, |
| "input_conv.weight": { |
| "w_norm": 7.95150089263916, |
| "g_norm": 0.06437147408723831, |
| "shape": [ |
| 192, |
| 6, |
| 3, |
| 3 |
| ] |
| }, |
| "input_conv.bias": { |
| "w_norm": 1.1900471448898315, |
| "g_norm": 0.0870174691081047, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.weight": { |
| "w_norm": 13.795973777770996, |
| "g_norm": 0.022462671622633934, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.bias": { |
| "w_norm": 0.10943365097045898, |
| "g_norm": 0.03537271171808243, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv1.weight": { |
| "w_norm": 8.521758079528809, |
| "g_norm": 0.18474127352237701, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv1.bias": { |
| "w_norm": 0.20209793746471405, |
| "g_norm": 0.019622452557086945, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.t_proj.weight": { |
| "w_norm": 8.095259666442871, |
| "g_norm": 0.009041828103363514, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.0.t_proj.bias": { |
| "w_norm": 0.5026264190673828, |
| "g_norm": 0.019634367898106575, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.weight": { |
| "w_norm": 13.82917308807373, |
| "g_norm": 0.007581069599837065, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.bias": { |
| "w_norm": 0.07328630983829498, |
| "g_norm": 0.008389967493712902, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv2.weight": { |
| "w_norm": 8.377623558044434, |
| "g_norm": 0.05995709449052811, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv2.bias": { |
| "w_norm": 0.18875400722026825, |
| "g_norm": 0.004590434953570366, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.weight": { |
| "w_norm": 13.848093032836914, |
| "g_norm": 0.00014158007979858667, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.bias": { |
| "w_norm": 0.042593494057655334, |
| "g_norm": 0.0001738002902129665, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv1.weight": { |
| "w_norm": 8.286897659301758, |
| "g_norm": 0.0024213152937591076, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv1.bias": { |
| "w_norm": 0.19198547303676605, |
| "g_norm": 0.0001342037576250732, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.t_proj.weight": { |
| "w_norm": 8.197196960449219, |
| "g_norm": 0.00010108872811542824, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.1.t_proj.bias": { |
| "w_norm": 0.5223928093910217, |
| "g_norm": 0.00013416219735518098, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.weight": { |
| "w_norm": 13.820600509643555, |
| "g_norm": 0.00015041811275295913, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.bias": { |
| "w_norm": 0.06845534592866898, |
| "g_norm": 0.0001458735205233097, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv2.weight": { |
| "w_norm": 8.267409324645996, |
| "g_norm": 0.002303261077031493, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv2.bias": { |
| "w_norm": 0.2089850902557373, |
| "g_norm": 0.00011512059427332133, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.weight": { |
| "w_norm": 13.81585693359375, |
| "g_norm": 2.5233930500689894e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.bias": { |
| "w_norm": 0.06356502324342728, |
| "g_norm": 2.1722678866353817e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.conv1.weight": { |
| "w_norm": 11.720547676086426, |
| "g_norm": 0.0006722903344780207, |
| "shape": [ |
| 384, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv1.bias": { |
| "w_norm": 0.26672983169555664, |
| "g_norm": 2.4164550268324092e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.t_proj.weight": { |
| "w_norm": 11.856393814086914, |
| "g_norm": 2.154646244889591e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.2.t_proj.bias": { |
| "w_norm": 0.7216393947601318, |
| "g_norm": 2.4155662686098367e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.weight": { |
| "w_norm": 19.427146911621094, |
| "g_norm": 6.073198892408982e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.bias": { |
| "w_norm": 0.21375973522663116, |
| "g_norm": 3.191018186043948e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.conv2.weight": { |
| "w_norm": 11.735209465026855, |
| "g_norm": 0.0007884344086050987, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv2.bias": { |
| "w_norm": 0.20197713375091553, |
| "g_norm": 2.198413130827248e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.skip.weight": { |
| "w_norm": 11.192889213562012, |
| "g_norm": 0.0007470609270967543, |
| "shape": [ |
| 384, |
| 192, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.2.skip.bias": { |
| "w_norm": 0.8284077644348145, |
| "g_norm": 2.198413130827248e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.weight": { |
| "w_norm": 19.583465576171875, |
| "g_norm": 4.372146577225067e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.bias": { |
| "w_norm": 0.05683818459510803, |
| "g_norm": 3.3718665690685157e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv1.weight": { |
| "w_norm": 11.894614219665527, |
| "g_norm": 7.07660146872513e-05, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv1.bias": { |
| "w_norm": 0.1975417584180832, |
| "g_norm": 2.109597971866606e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.t_proj.weight": { |
| "w_norm": 11.404732704162598, |
| "g_norm": 1.5137313766899751e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.3.t_proj.bias": { |
| "w_norm": 0.6922511458396912, |
| "g_norm": 2.109284650941845e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.weight": { |
| "w_norm": 19.406980514526367, |
| "g_norm": 3.6609660583053483e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.bias": { |
| "w_norm": 0.19815607368946075, |
| "g_norm": 3.5790540096058976e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv2.weight": { |
| "w_norm": 11.528669357299805, |
| "g_norm": 0.00015909784997347742, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv2.bias": { |
| "w_norm": 0.19179506599903107, |
| "g_norm": 5.194372533878777e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.weight": { |
| "w_norm": 19.590572357177734, |
| "g_norm": 1.2455842579583987e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.bias": { |
| "w_norm": 0.06201266124844551, |
| "g_norm": 1.6669519027345814e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.conv1.weight": { |
| "w_norm": 16.711240768432617, |
| "g_norm": 2.725605918385554e-05, |
| "shape": [ |
| 768, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv1.bias": { |
| "w_norm": 0.2846062183380127, |
| "g_norm": 1.3469352779793553e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.t_proj.weight": { |
| "w_norm": 16.05572509765625, |
| "g_norm": 7.266082775458926e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.4.t_proj.bias": { |
| "w_norm": 0.9931569695472717, |
| "g_norm": 1.3468946917782887e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.weight": { |
| "w_norm": 27.43099021911621, |
| "g_norm": 1.147222178587981e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.bias": { |
| "w_norm": 0.3143310844898224, |
| "g_norm": 2.341718982279417e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.conv2.weight": { |
| "w_norm": 16.32482147216797, |
| "g_norm": 6.44704487058334e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv2.bias": { |
| "w_norm": 0.1955537497997284, |
| "g_norm": 3.4157737900386564e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.skip.weight": { |
| "w_norm": 15.805265426635742, |
| "g_norm": 7.485902460757643e-05, |
| "shape": [ |
| 768, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.4.skip.bias": { |
| "w_norm": 0.8076781630516052, |
| "g_norm": 3.4157737900386564e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.weight": { |
| "w_norm": 27.713409423828125, |
| "g_norm": 3.229412470773241e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.bias": { |
| "w_norm": 0.05912580341100693, |
| "g_norm": 2.589934808838734e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv1.weight": { |
| "w_norm": 16.959115982055664, |
| "g_norm": 1.2284299373277463e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv1.bias": { |
| "w_norm": 0.20442481338977814, |
| "g_norm": 2.93640937343298e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.t_proj.weight": { |
| "w_norm": 16.0108642578125, |
| "g_norm": 1.805583877967365e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.5.t_proj.bias": { |
| "w_norm": 0.9978851079940796, |
| "g_norm": 2.9374672294579796e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.weight": { |
| "w_norm": 27.502161026000977, |
| "g_norm": 4.7526637558803486e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.bias": { |
| "w_norm": 0.24743936955928802, |
| "g_norm": 5.183755433790793e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv2.weight": { |
| "w_norm": 16.4858341217041, |
| "g_norm": 3.6588418879546225e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv2.bias": { |
| "w_norm": 0.1915251463651657, |
| "g_norm": 1.2130102504670504e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.3.norm.weight": { |
| "w_norm": 19.52772331237793, |
| "g_norm": 1.2919706477987347e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.norm.bias": { |
| "w_norm": 0.04880025237798691, |
| "g_norm": 2.042158257609117e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.qkv.weight": { |
| "w_norm": 19.67283821105957, |
| "g_norm": 3.667322744149715e-05, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.qkv.bias": { |
| "w_norm": 1.007038950920105, |
| "g_norm": 3.1300469345296733e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "enc_attns.3.proj.weight": { |
| "w_norm": 11.307280540466309, |
| "g_norm": 3.458759965724312e-05, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.proj.bias": { |
| "w_norm": 0.562714159488678, |
| "g_norm": 5.2210789363016374e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.4.norm.weight": { |
| "w_norm": 27.598068237304688, |
| "g_norm": 1.305578621213499e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.norm.bias": { |
| "w_norm": 0.051544707268476486, |
| "g_norm": 1.4329129953694064e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.qkv.weight": { |
| "w_norm": 27.992589950561523, |
| "g_norm": 4.070655631949194e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.qkv.bias": { |
| "w_norm": 1.0139776468276978, |
| "g_norm": 1.8191481103713159e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.4.proj.weight": { |
| "w_norm": 16.01902961730957, |
| "g_norm": 4.0771319618215784e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.proj.bias": { |
| "w_norm": 0.5725747346878052, |
| "g_norm": 3.0066312319831923e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.weight": { |
| "w_norm": 27.660709381103516, |
| "g_norm": 3.203698213383177e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.bias": { |
| "w_norm": 0.048515934497117996, |
| "g_norm": 4.2762215457514685e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.qkv.weight": { |
| "w_norm": 28.006317138671875, |
| "g_norm": 1.4664356058347039e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.qkv.bias": { |
| "w_norm": 0.9999328255653381, |
| "g_norm": 6.966732257751573e-07, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.5.proj.weight": { |
| "w_norm": 16.06597137451172, |
| "g_norm": 1.613639142306056e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.proj.bias": { |
| "w_norm": 0.5973749160766602, |
| "g_norm": 1.1837975080197793e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_downs.0.op.weight": { |
| "w_norm": 8.197033882141113, |
| "g_norm": 0.00462199654430151, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.0.op.bias": { |
| "w_norm": 0.1845771074295044, |
| "g_norm": 0.00018332099716644734, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.1.op.weight": { |
| "w_norm": 8.111349105834961, |
| "g_norm": 0.001202525570988655, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.1.op.bias": { |
| "w_norm": 0.19379548728466034, |
| "g_norm": 1.966803210962098e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.2.op.weight": { |
| "w_norm": 11.510343551635742, |
| "g_norm": 0.00029774088761769235, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.2.op.bias": { |
| "w_norm": 0.1884956955909729, |
| "g_norm": 4.963231731380802e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.3.op.weight": { |
| "w_norm": 11.436445236206055, |
| "g_norm": 0.00012070700904587284, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.3.op.bias": { |
| "w_norm": 0.1916549801826477, |
| "g_norm": 2.680138095456641e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.4.op.weight": { |
| "w_norm": 16.354839324951172, |
| "g_norm": 6.465298793045804e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.4.op.bias": { |
| "w_norm": 0.19893445074558258, |
| "g_norm": 1.2337447969912319e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.weight": { |
| "w_norm": 27.714229583740234, |
| "g_norm": 2.476989209299063e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.bias": { |
| "w_norm": 0.05086077004671097, |
| "g_norm": 2.4525201069991454e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv1.weight": { |
| "w_norm": 16.912973403930664, |
| "g_norm": 8.582318514527287e-06, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv1.bias": { |
| "w_norm": 0.2083008736371994, |
| "g_norm": 2.025992245080488e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.t_proj.weight": { |
| "w_norm": 16.017248153686523, |
| "g_norm": 1.0194400346108523e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid1.t_proj.bias": { |
| "w_norm": 1.0305330753326416, |
| "g_norm": 2.0257930088973808e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.weight": { |
| "w_norm": 27.515790939331055, |
| "g_norm": 2.3792208025952277e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.bias": { |
| "w_norm": 0.2597298324108124, |
| "g_norm": 3.9932524487085175e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv2.weight": { |
| "w_norm": 16.51875877380371, |
| "g_norm": 1.753577816998586e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv2.bias": { |
| "w_norm": 0.19382815062999725, |
| "g_norm": 7.970655815370264e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.weight": { |
| "w_norm": 27.66170883178711, |
| "g_norm": 2.1583025500149233e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.bias": { |
| "w_norm": 0.05416889116168022, |
| "g_norm": 2.994936494360445e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.qkv.weight": { |
| "w_norm": 27.94774627685547, |
| "g_norm": 9.899607903207652e-06, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.qkv.bias": { |
| "w_norm": 0.9840787053108215, |
| "g_norm": 4.803904403161141e-07, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "mid_attn.proj.weight": { |
| "w_norm": 16.05330467224121, |
| "g_norm": 1.0125810149475e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.proj.bias": { |
| "w_norm": 0.5761896371841431, |
| "g_norm": 7.825606189726386e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.weight": { |
| "w_norm": 27.72232437133789, |
| "g_norm": 2.2575868285912293e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.bias": { |
| "w_norm": 0.06341087073087692, |
| "g_norm": 1.9701586495557422e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv1.weight": { |
| "w_norm": 17.024290084838867, |
| "g_norm": 7.21612968845875e-06, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv1.bias": { |
| "w_norm": 0.20101290941238403, |
| "g_norm": 1.657937644949925e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.t_proj.weight": { |
| "w_norm": 16.00647735595703, |
| "g_norm": 1.1699727764380441e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid2.t_proj.bias": { |
| "w_norm": 0.9998338222503662, |
| "g_norm": 1.6574250594203477e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.weight": { |
| "w_norm": 27.492137908935547, |
| "g_norm": 1.739176553883226e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.bias": { |
| "w_norm": 0.26778048276901245, |
| "g_norm": 3.6507077538772137e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv2.weight": { |
| "w_norm": 16.478769302368164, |
| "g_norm": 1.3940637472842354e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv2.bias": { |
| "w_norm": 0.1897919923067093, |
| "g_norm": 7.798574301887129e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm1.weight": { |
| "w_norm": 39.217708587646484, |
| "g_norm": 3.7901781979599036e-07, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.norm1.bias": { |
| "w_norm": 0.09732687473297119, |
| "g_norm": 3.5458742786431685e-07, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.conv1.weight": { |
| "w_norm": 17.912757873535156, |
| "g_norm": 1.0468619620951358e-05, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv1.bias": { |
| "w_norm": 0.15449513494968414, |
| "g_norm": 1.8222155517833016e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.t_proj.weight": { |
| "w_norm": 16.024795532226562, |
| "g_norm": 1.1644377195807465e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.0.t_proj.bias": { |
| "w_norm": 1.0251551866531372, |
| "g_norm": 1.8221867037482298e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.weight": { |
| "w_norm": 27.449193954467773, |
| "g_norm": 3.468843203791039e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.bias": { |
| "w_norm": 0.3095521032810211, |
| "g_norm": 8.821044730211725e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.conv2.weight": { |
| "w_norm": 16.268291473388672, |
| "g_norm": 2.5844508854788728e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv2.bias": { |
| "w_norm": 0.19935624301433563, |
| "g_norm": 1.6814665286801755e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.skip.weight": { |
| "w_norm": 15.761346817016602, |
| "g_norm": 9.896884148474783e-05, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.0.skip.bias": { |
| "w_norm": 0.410925954580307, |
| "g_norm": 1.6814665286801755e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm1.weight": { |
| "w_norm": 39.21905517578125, |
| "g_norm": 6.33988861409307e-07, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.norm1.bias": { |
| "w_norm": 0.09087496995925903, |
| "g_norm": 5.014872499486955e-07, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.conv1.weight": { |
| "w_norm": 17.78136444091797, |
| "g_norm": 2.920708357123658e-05, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv1.bias": { |
| "w_norm": 0.14948569238185883, |
| "g_norm": 4.78054118957516e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.t_proj.weight": { |
| "w_norm": 16.056991577148438, |
| "g_norm": 4.6630319161522493e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.1.t_proj.bias": { |
| "w_norm": 0.9883320927619934, |
| "g_norm": 4.781649636242946e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.weight": { |
| "w_norm": 27.447290420532227, |
| "g_norm": 9.171581609734858e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.bias": { |
| "w_norm": 0.3388512134552002, |
| "g_norm": 2.0488243990257615e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.conv2.weight": { |
| "w_norm": 16.2231502532959, |
| "g_norm": 8.360083302250132e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv2.bias": { |
| "w_norm": 0.19709675014019012, |
| "g_norm": 5.03153023601044e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.skip.weight": { |
| "w_norm": 15.74288272857666, |
| "g_norm": 0.00020004258840344846, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.1.skip.bias": { |
| "w_norm": 0.4062535762786865, |
| "g_norm": 5.03153023601044e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.2.norm1.weight": { |
| "w_norm": 33.958805084228516, |
| "g_norm": 1.8269259953740402e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.norm1.bias": { |
| "w_norm": 0.09479564428329468, |
| "g_norm": 1.4436129731620895e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.conv1.weight": { |
| "w_norm": 12.608702659606934, |
| "g_norm": 8.076793164946139e-05, |
| "shape": [ |
| 384, |
| 1152, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv1.bias": { |
| "w_norm": 0.1214609444141388, |
| "g_norm": 1.3795322502119234e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.t_proj.weight": { |
| "w_norm": 11.346697807312012, |
| "g_norm": 1.288618136641162e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.2.t_proj.bias": { |
| "w_norm": 0.7093044519424438, |
| "g_norm": 1.3801052318740403e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.weight": { |
| "w_norm": 19.435365676879883, |
| "g_norm": 3.010250793522573e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.bias": { |
| "w_norm": 0.22179819643497467, |
| "g_norm": 5.038673407398164e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.conv2.weight": { |
| "w_norm": 11.456557273864746, |
| "g_norm": 0.00020378570479806513, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv2.bias": { |
| "w_norm": 0.1886361539363861, |
| "g_norm": 1.2030825928377453e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.skip.weight": { |
| "w_norm": 10.977109909057617, |
| "g_norm": 0.0004246227617841214, |
| "shape": [ |
| 384, |
| 1152, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.2.skip.bias": { |
| "w_norm": 0.331463485956192, |
| "g_norm": 1.2030825928377453e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm1.weight": { |
| "w_norm": 27.710662841796875, |
| "g_norm": 1.555925700813532e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.norm1.bias": { |
| "w_norm": 0.08933005481958389, |
| "g_norm": 1.3325834515853785e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.conv1.weight": { |
| "w_norm": 12.44627571105957, |
| "g_norm": 0.0004514533211477101, |
| "shape": [ |
| 384, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv1.bias": { |
| "w_norm": 0.14480382204055786, |
| "g_norm": 8.666344001539983e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.t_proj.weight": { |
| "w_norm": 11.391429901123047, |
| "g_norm": 9.20170441531809e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.3.t_proj.bias": { |
| "w_norm": 0.7092675566673279, |
| "g_norm": 8.667771908221766e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.weight": { |
| "w_norm": 19.412303924560547, |
| "g_norm": 2.933637188107241e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.bias": { |
| "w_norm": 0.22528940439224243, |
| "g_norm": 2.264487739012111e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.conv2.weight": { |
| "w_norm": 11.52460765838623, |
| "g_norm": 0.0008822944364510477, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv2.bias": { |
| "w_norm": 0.1936788111925125, |
| "g_norm": 4.2916828533634543e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.skip.weight": { |
| "w_norm": 11.064810752868652, |
| "g_norm": 0.0010949639836326241, |
| "shape": [ |
| 384, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.3.skip.bias": { |
| "w_norm": 0.4060087203979492, |
| "g_norm": 4.2916828533634543e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.4.norm1.weight": { |
| "w_norm": 23.974384307861328, |
| "g_norm": 0.00010767428466351703, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.norm1.bias": { |
| "w_norm": 0.09818897396326065, |
| "g_norm": 0.00010226306039839983, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.conv1.weight": { |
| "w_norm": 8.616304397583008, |
| "g_norm": 0.002126651583239436, |
| "shape": [ |
| 192, |
| 576, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv1.bias": { |
| "w_norm": 0.11683345586061478, |
| "g_norm": 6.0171176301082596e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.t_proj.weight": { |
| "w_norm": 8.06219482421875, |
| "g_norm": 6.534672138513997e-05, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.4.t_proj.bias": { |
| "w_norm": 0.4816136360168457, |
| "g_norm": 6.018929343554191e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.weight": { |
| "w_norm": 13.791584968566895, |
| "g_norm": 0.00013399690215010196, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.bias": { |
| "w_norm": 0.08628630638122559, |
| "g_norm": 0.00014492537593469024, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.conv2.weight": { |
| "w_norm": 8.11715030670166, |
| "g_norm": 0.00321268942207098, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv2.bias": { |
| "w_norm": 0.19586454331874847, |
| "g_norm": 0.0001909145648824051, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.skip.weight": { |
| "w_norm": 7.841552734375, |
| "g_norm": 0.004247691016644239, |
| "shape": [ |
| 192, |
| 576, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.4.skip.bias": { |
| "w_norm": 0.3465918302536011, |
| "g_norm": 0.0001909145648824051, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm1.weight": { |
| "w_norm": 19.53300666809082, |
| "g_norm": 0.0025150158908218145, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.norm1.bias": { |
| "w_norm": 0.12533682584762573, |
| "g_norm": 0.0028404968325048685, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.conv1.weight": { |
| "w_norm": 8.668478965759277, |
| "g_norm": 0.028973735868930817, |
| "shape": [ |
| 192, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv1.bias": { |
| "w_norm": 0.13875038921833038, |
| "g_norm": 0.0014434006297960877, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.t_proj.weight": { |
| "w_norm": 8.21015453338623, |
| "g_norm": 0.0008499801042489707, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.5.t_proj.bias": { |
| "w_norm": 0.5054141283035278, |
| "g_norm": 0.001443629153072834, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.weight": { |
| "w_norm": 13.897296905517578, |
| "g_norm": 0.002029021503403783, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.bias": { |
| "w_norm": 0.045003172010183334, |
| "g_norm": 0.0019661581609398127, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.conv2.weight": { |
| "w_norm": 8.435464859008789, |
| "g_norm": 0.029161185026168823, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv2.bias": { |
| "w_norm": 0.20200084149837494, |
| "g_norm": 0.0014317930908873677, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.skip.weight": { |
| "w_norm": 8.107450485229492, |
| "g_norm": 0.022503562271595, |
| "shape": [ |
| 192, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.5.skip.bias": { |
| "w_norm": 0.4229964017868042, |
| "g_norm": 0.0014317930908873677, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_attns.0.norm.weight": { |
| "w_norm": 27.606098175048828, |
| "g_norm": 1.4698603081342299e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.norm.bias": { |
| "w_norm": 0.05232677236199379, |
| "g_norm": 1.6636099644529168e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.qkv.weight": { |
| "w_norm": 28.047964096069336, |
| "g_norm": 3.253455724916421e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.qkv.bias": { |
| "w_norm": 1.007122278213501, |
| "g_norm": 1.4771649148315191e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.0.proj.weight": { |
| "w_norm": 16.108665466308594, |
| "g_norm": 2.8333268346614204e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.proj.bias": { |
| "w_norm": 0.5780074596405029, |
| "g_norm": 1.8256989733345108e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.weight": { |
| "w_norm": 27.601665496826172, |
| "g_norm": 2.4182666038541356e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.bias": { |
| "w_norm": 0.049121566116809845, |
| "g_norm": 2.8192550871608546e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.qkv.weight": { |
| "w_norm": 28.07448387145996, |
| "g_norm": 7.991576421773061e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.qkv.bias": { |
| "w_norm": 1.0039130449295044, |
| "g_norm": 3.495277724141488e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.1.proj.weight": { |
| "w_norm": 16.064023971557617, |
| "g_norm": 6.958303129067644e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.proj.bias": { |
| "w_norm": 0.5789670944213867, |
| "g_norm": 5.107773631607415e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.2.norm.weight": { |
| "w_norm": 19.495100021362305, |
| "g_norm": 3.4630834306881297e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.norm.bias": { |
| "w_norm": 0.026043370366096497, |
| "g_norm": 5.35019989911234e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.qkv.weight": { |
| "w_norm": 19.710468292236328, |
| "g_norm": 0.00010689659393392503, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.qkv.bias": { |
| "w_norm": 0.98268061876297, |
| "g_norm": 7.844309038773645e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_attns.2.proj.weight": { |
| "w_norm": 11.305863380432129, |
| "g_norm": 0.00010106317495228723, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.proj.bias": { |
| "w_norm": 0.5879626274108887, |
| "g_norm": 1.2567536941787694e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.0.op.weight": { |
| "w_norm": 16.159915924072266, |
| "g_norm": 0.00017089844914153218, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.0.op.bias": { |
| "w_norm": 0.20129258930683136, |
| "g_norm": 2.0403990674822126e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.1.op.weight": { |
| "w_norm": 16.166606903076172, |
| "g_norm": 0.00038573183701373637, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.1.op.bias": { |
| "w_norm": 0.1936124861240387, |
| "g_norm": 5.565851552091772e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.2.op.weight": { |
| "w_norm": 11.31502914428711, |
| "g_norm": 0.0009087788057513535, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.2.op.bias": { |
| "w_norm": 0.197477325797081, |
| "g_norm": 1.888297265395522e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.3.op.weight": { |
| "w_norm": 11.352828979492188, |
| "g_norm": 0.002423783065751195, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.3.op.bias": { |
| "w_norm": 0.19884906709194183, |
| "g_norm": 6.680266233161092e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.4.op.weight": { |
| "w_norm": 8.010747909545898, |
| "g_norm": 0.010360135696828365, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.4.op.bias": { |
| "w_norm": 0.19764398038387299, |
| "g_norm": 0.00024076507543213665, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.weight": { |
| "w_norm": 13.761808395385742, |
| "g_norm": 0.003240334102883935, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.bias": { |
| "w_norm": 0.11472446471452713, |
| "g_norm": 0.004786268807947636, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_conv.weight": { |
| "w_norm": 0.15211158990859985, |
| "g_norm": 0.5555833578109741, |
| "shape": [ |
| 3, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "out_conv.bias": { |
| "w_norm": 0.009061155840754509, |
| "g_norm": 0.03287523612380028, |
| "shape": [ |
| 3 |
| ] |
| } |
| }, |
| "activations": { |
| "input_conv": { |
| "mean": 0.00040025782072916627, |
| "std": 0.3693671226501465, |
| "abs_max": 2.671875, |
| "frac_dead": 0.00047539008664898574, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.1": { |
| "mean": 0.010746731422841549, |
| "std": 0.2820868194103241, |
| "abs_max": 4.25, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.3": { |
| "mean": 0.02136988565325737, |
| "std": 0.21643811464309692, |
| "abs_max": 1.0625, |
| "frac_dead": 0.00016276042151730508, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm1": { |
| "mean": -0.006252565421164036, |
| "std": 0.9951425790786743, |
| "abs_max": 6.74277925491333, |
| "frac_dead": 4.76837158203125e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv1": { |
| "mean": -0.041000474244356155, |
| "std": 0.6137309670448303, |
| "abs_max": 10.75, |
| "frac_dead": 6.946259236428887e-05, |
| "frac_sat": 1.0265244299034748e-07 |
| }, |
| "enc_blocks.0.t_proj": { |
| "mean": 0.0007395787979476154, |
| "std": 0.2251749038696289, |
| "abs_max": 2.40625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm2": { |
| "mean": -0.004515793640166521, |
| "std": 0.9990665316581726, |
| "abs_max": 12.43277359008789, |
| "frac_dead": 1.3841523696100921e-06, |
| "frac_sat": 4.801485147254425e-07 |
| }, |
| "enc_blocks.0.conv2": { |
| "mean": -0.030521171167492867, |
| "std": 0.8776020407676697, |
| "abs_max": 8.75, |
| "frac_dead": 3.275275230407715e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.0.op": { |
| "mean": -0.035549554973840714, |
| "std": 1.359336495399475, |
| "abs_max": 13.75, |
| "frac_dead": 2.3351774871116504e-05, |
| "frac_sat": 3.378921246621758e-05 |
| }, |
| "enc_blocks.1.norm1": { |
| "mean": -0.0025818953290581703, |
| "std": 0.9994575381278992, |
| "abs_max": 9.819805145263672, |
| "frac_dead": 9.5367431640625e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv1": { |
| "mean": -0.2658883333206177, |
| "std": 1.0592257976531982, |
| "abs_max": 12.5, |
| "frac_dead": 2.8875139832962304e-05, |
| "frac_sat": 1.3642841167893494e-06 |
| }, |
| "enc_blocks.1.t_proj": { |
| "mean": 0.0059111700393259525, |
| "std": 0.24908672273159027, |
| "abs_max": 1.9453125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.norm2": { |
| "mean": -0.005134095437824726, |
| "std": 0.9981088042259216, |
| "abs_max": 12.5274658203125, |
| "frac_dead": 6.22537413619284e-07, |
| "frac_sat": 6.490283794846619e-07 |
| }, |
| "enc_blocks.1.conv2": { |
| "mean": 0.02617700956761837, |
| "std": 0.9424678087234497, |
| "abs_max": 10.875, |
| "frac_dead": 4.083580643055029e-05, |
| "frac_sat": 2.2517311037972831e-07 |
| }, |
| "enc_downs.1.op": { |
| "mean": -0.0690908432006836, |
| "std": 1.943811058998108, |
| "abs_max": 14.0625, |
| "frac_dead": 1.811981201171875e-05, |
| "frac_sat": 4.725986218545586e-05 |
| }, |
| "enc_blocks.2.norm1": { |
| "mean": -0.0037061066832393408, |
| "std": 0.9978165626525879, |
| "abs_max": 10.55722713470459, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 5.298190686175985e-08 |
| }, |
| "enc_blocks.2.conv1": { |
| "mean": -0.13071851432323456, |
| "std": 1.3800336122512817, |
| "abs_max": 22.375, |
| "frac_dead": 3.655751424957998e-05, |
| "frac_sat": 0.00047132704639807343 |
| }, |
| "enc_blocks.2.t_proj": { |
| "mean": -0.0055412352085113525, |
| "std": 0.332256555557251, |
| "abs_max": 3.015625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.norm2": { |
| "mean": -0.011643648147583008, |
| "std": 0.9939536452293396, |
| "abs_max": 12.937939643859863, |
| "frac_dead": 1.2185838613731903e-06, |
| "frac_sat": 1.3589859008789062e-05 |
| }, |
| "enc_blocks.2.conv2": { |
| "mean": -0.0342453196644783, |
| "std": 0.7672107219696045, |
| "abs_max": 8.8125, |
| "frac_dead": 2.961688551295083e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.skip": { |
| "mean": -0.037515006959438324, |
| "std": 1.0088363885879517, |
| "abs_max": 9.625, |
| "frac_dead": 0.00010739432764239609, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.2.op": { |
| "mean": -0.199848935008049, |
| "std": 1.5087478160858154, |
| "abs_max": 11.75, |
| "frac_dead": 1.4623005881730933e-05, |
| "frac_sat": 2.1192763597355224e-06 |
| }, |
| "enc_blocks.3.norm1": { |
| "mean": -0.00175758870318532, |
| "std": 0.99953293800354, |
| "abs_max": 8.614314079284668, |
| "frac_dead": 1.0596381798677612e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.conv1": { |
| "mean": -0.10498303920030594, |
| "std": 1.746254563331604, |
| "abs_max": 30.75, |
| "frac_dead": 2.1616617232211865e-05, |
| "frac_sat": 0.002836121479049325 |
| }, |
| "enc_blocks.3.t_proj": { |
| "mean": 0.002910944866016507, |
| "std": 0.14892221987247467, |
| "abs_max": 1.1875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.norm2": { |
| "mean": -0.01072887796908617, |
| "std": 0.9917898774147034, |
| "abs_max": 16.629051208496094, |
| "frac_dead": 3.602769766075653e-06, |
| "frac_sat": 6.114112329669297e-05 |
| }, |
| "enc_blocks.3.conv2": { |
| "mean": -0.02169557847082615, |
| "std": 0.7376501560211182, |
| "abs_max": 8.375, |
| "frac_dead": 3.549787652445957e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.3.norm": { |
| "mean": 0.0005119277630001307, |
| "std": 0.9960955381393433, |
| "abs_max": 8.947876930236816, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.3.qkv": { |
| "mean": -0.00030917784897610545, |
| "std": 0.6075136661529541, |
| "abs_max": 5.34375, |
| "frac_dead": 0.00010931933502433822, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.3.proj": { |
| "mean": 0.007649092935025692, |
| "std": 0.26445072889328003, |
| "abs_max": 1.3046875, |
| "frac_dead": 0.000232696533203125, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.3.op": { |
| "mean": -0.23308219015598297, |
| "std": 1.8356003761291504, |
| "abs_max": 16.875, |
| "frac_dead": 1.5682644516346045e-05, |
| "frac_sat": 0.00038316514110192657 |
| }, |
| "enc_blocks.4.norm1": { |
| "mean": -0.0006192077999003232, |
| "std": 1.0010056495666504, |
| "abs_max": 11.152945518493652, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 4.238552548940788e-07 |
| }, |
| "enc_blocks.4.conv1": { |
| "mean": -0.2928019165992737, |
| "std": 1.740490198135376, |
| "abs_max": 31.125, |
| "frac_dead": 2.8822158128605224e-05, |
| "frac_sat": 0.0039316811598837376 |
| }, |
| "enc_blocks.4.t_proj": { |
| "mean": -0.0017857596976682544, |
| "std": 0.10373267531394958, |
| "abs_max": 0.76171875, |
| "frac_dead": 5.425347262644209e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.norm2": { |
| "mean": -0.011213724501430988, |
| "std": 0.9912196397781372, |
| "abs_max": 16.70511245727539, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.00014156765246298164 |
| }, |
| "enc_blocks.4.conv2": { |
| "mean": 0.031458597630262375, |
| "std": 0.4912275969982147, |
| "abs_max": 3.84375, |
| "frac_dead": 3.369649130036123e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.skip": { |
| "mean": -0.03792019560933113, |
| "std": 0.958321213722229, |
| "abs_max": 7.15625, |
| "frac_dead": 8.138021075865254e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.norm": { |
| "mean": 9.616216266294941e-05, |
| "std": 0.9960263967514038, |
| "abs_max": 8.730345726013184, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.qkv": { |
| "mean": -0.006408622022718191, |
| "std": 0.685590386390686, |
| "abs_max": 6.5, |
| "frac_dead": 6.696912896586582e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.proj": { |
| "mean": 0.003732851007953286, |
| "std": 0.4339083433151245, |
| "abs_max": 3.515625, |
| "frac_dead": 0.00012736850476358086, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.4.op": { |
| "mean": 0.0982859805226326, |
| "std": 1.6599680185317993, |
| "abs_max": 12.25, |
| "frac_dead": 1.4411079064302612e-05, |
| "frac_sat": 7.968478894326836e-05 |
| }, |
| "enc_blocks.5.norm1": { |
| "mean": 2.0242399841663428e-05, |
| "std": 1.0001006126403809, |
| "abs_max": 5.786524295806885, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv1": { |
| "mean": -0.39854559302330017, |
| "std": 1.6837373971939087, |
| "abs_max": 14.8125, |
| "frac_dead": 1.610649997019209e-05, |
| "frac_sat": 0.0010392931289970875 |
| }, |
| "enc_blocks.5.t_proj": { |
| "mean": -0.0011938775423914194, |
| "std": 0.09263734519481659, |
| "abs_max": 0.77734375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.norm2": { |
| "mean": -0.008912770077586174, |
| "std": 0.9928245544433594, |
| "abs_max": 8.65130615234375, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv2": { |
| "mean": 0.19011344015598297, |
| "std": 1.201659917831421, |
| "abs_max": 7.6875, |
| "frac_dead": 7.62939453125e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.norm": { |
| "mean": 0.00032190897036343813, |
| "std": 0.9978728294372559, |
| "abs_max": 5.934600353240967, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.qkv": { |
| "mean": -0.015497934073209763, |
| "std": 0.8438277840614319, |
| "abs_max": 5.46875, |
| "frac_dead": 5.849202352692373e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.proj": { |
| "mean": -0.0038925441913306713, |
| "std": 0.32418742775917053, |
| "abs_max": 2.453125, |
| "frac_dead": 0.00017038981604855508, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm1": { |
| "mean": 0.0001421187917003408, |
| "std": 1.0004252195358276, |
| "abs_max": 5.862927436828613, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv1": { |
| "mean": -0.8414751291275024, |
| "std": 2.068591833114624, |
| "abs_max": 20.375, |
| "frac_dead": 2.1192763597355224e-05, |
| "frac_sat": 0.0095986258238554 |
| }, |
| "mid1.t_proj": { |
| "mean": 0.0026207775808870792, |
| "std": 0.09048116207122803, |
| "abs_max": 0.62109375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm2": { |
| "mean": -0.009563598781824112, |
| "std": 0.994515061378479, |
| "abs_max": 8.810402870178223, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv2": { |
| "mean": 0.041060809046030045, |
| "std": 0.7921715974807739, |
| "abs_max": 4.96875, |
| "frac_dead": 2.8822158128605224e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.norm": { |
| "mean": 0.0004211680789012462, |
| "std": 0.9976292252540588, |
| "abs_max": 7.214486122131348, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.qkv": { |
| "mean": 0.005006986204534769, |
| "std": 0.7217532992362976, |
| "abs_max": 4.625, |
| "frac_dead": 6.0470018070191145e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.proj": { |
| "mean": -0.010449005290865898, |
| "std": 0.3259406089782715, |
| "abs_max": 2.15625, |
| "frac_dead": 0.00012885199976153672, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm1": { |
| "mean": 0.000711227476131171, |
| "std": 1.0010067224502563, |
| "abs_max": 7.27166748046875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv1": { |
| "mean": -0.9361047744750977, |
| "std": 2.149503707885742, |
| "abs_max": 21.375, |
| "frac_dead": 2.6279025405528955e-05, |
| "frac_sat": 0.014512804336845875 |
| }, |
| "mid2.t_proj": { |
| "mean": 0.0011170649668201804, |
| "std": 0.09452700614929199, |
| "abs_max": 0.7109375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm2": { |
| "mean": -0.010762469843029976, |
| "std": 0.9965609312057495, |
| "abs_max": 9.152743339538574, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv2": { |
| "mean": 0.007028906140476465, |
| "std": 0.4812912940979004, |
| "abs_max": 2.875, |
| "frac_dead": 2.7126736313221045e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm1": { |
| "mean": 0.0011261552572250366, |
| "std": 1.001523494720459, |
| "abs_max": 6.811408519744873, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.conv1": { |
| "mean": -2.7249886989593506, |
| "std": 4.642963886260986, |
| "abs_max": 51.25, |
| "frac_dead": 4.238552719471045e-06, |
| "frac_sat": 0.06366220861673355 |
| }, |
| "dec_blocks.0.t_proj": { |
| "mean": -0.0014680986059829593, |
| "std": 0.09673027694225311, |
| "abs_max": 0.6171875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm2": { |
| "mean": -0.01321282796561718, |
| "std": 0.9973823428153992, |
| "abs_max": 9.541091918945312, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.conv2": { |
| "mean": -0.00643331790342927, |
| "std": 0.37738314270973206, |
| "abs_max": 2.484375, |
| "frac_dead": 3.899468356394209e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.skip": { |
| "mean": 0.005534242372959852, |
| "std": 1.2259471416473389, |
| "abs_max": 7.625, |
| "frac_dead": 2.5431314497836865e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.norm": { |
| "mean": 7.251939678099006e-05, |
| "std": 0.9961729049682617, |
| "abs_max": 4.991007328033447, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.qkv": { |
| "mean": -0.032485075294971466, |
| "std": 0.6778285503387451, |
| "abs_max": 4.375, |
| "frac_dead": 6.668655987596139e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.proj": { |
| "mean": -0.00447096349671483, |
| "std": 0.5198463797569275, |
| "abs_max": 2.59375, |
| "frac_dead": 8.053249621298164e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.0.op": { |
| "mean": 0.11986494064331055, |
| "std": 1.3657621145248413, |
| "abs_max": 7.71875, |
| "frac_dead": 1.229180270456709e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.norm1": { |
| "mean": 0.0005290090339258313, |
| "std": 1.000602126121521, |
| "abs_max": 7.9642744064331055, |
| "frac_dead": 9.5367431640625e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.conv1": { |
| "mean": -2.2336854934692383, |
| "std": 3.7236015796661377, |
| "abs_max": 44.25, |
| "frac_dead": 6.145901352283545e-06, |
| "frac_sat": 0.04314698278903961 |
| }, |
| "dec_blocks.1.t_proj": { |
| "mean": -0.0013965911930426955, |
| "std": 0.09607820212841034, |
| "abs_max": 0.73046875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.norm2": { |
| "mean": -0.014699429273605347, |
| "std": 0.9981132745742798, |
| "abs_max": 9.811440467834473, |
| "frac_dead": 1.4834934063401306e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.conv2": { |
| "mean": -0.00016551630687899888, |
| "std": 0.3925391137599945, |
| "abs_max": 2.421875, |
| "frac_dead": 3.729926174855791e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.skip": { |
| "mean": -0.011327608488500118, |
| "std": 0.8946093916893005, |
| "abs_max": 5.21875, |
| "frac_dead": 4.217359673930332e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.norm": { |
| "mean": 7.759490108583122e-05, |
| "std": 0.9962604641914368, |
| "abs_max": 6.363803863525391, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.qkv": { |
| "mean": 0.01140259113162756, |
| "std": 0.6308833956718445, |
| "abs_max": 4.125, |
| "frac_dead": 7.212603668449447e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.proj": { |
| "mean": 0.009470382705330849, |
| "std": 0.3589651584625244, |
| "abs_max": 1.8203125, |
| "frac_dead": 0.0001373291015625, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.1.op": { |
| "mean": 0.05434713140130043, |
| "std": 0.9426736235618591, |
| "abs_max": 5.375, |
| "frac_dead": 1.7589993149158545e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm1": { |
| "mean": -0.00016650006000418216, |
| "std": 1.0008327960968018, |
| "abs_max": 9.631314277648926, |
| "frac_dead": 7.417467031700653e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.conv1": { |
| "mean": -1.0471075773239136, |
| "std": 3.5208213329315186, |
| "abs_max": 40.25, |
| "frac_dead": 9.112887710216455e-06, |
| "frac_sat": 0.041029613465070724 |
| }, |
| "dec_blocks.2.t_proj": { |
| "mean": -0.004079055972397327, |
| "std": 0.10816413164138794, |
| "abs_max": 0.6796875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm2": { |
| "mean": -0.014768723398447037, |
| "std": 1.0005854368209839, |
| "abs_max": 9.347208976745605, |
| "frac_dead": 1.1656019296424347e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.conv2": { |
| "mean": 0.023853028193116188, |
| "std": 0.3805305063724518, |
| "abs_max": 2.25, |
| "frac_dead": 5.361768853617832e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.skip": { |
| "mean": 0.032520074397325516, |
| "std": 0.8015388250350952, |
| "abs_max": 6.1875, |
| "frac_dead": 4.355112832854502e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.norm": { |
| "mean": -0.00016280759882647544, |
| "std": 0.9950010776519775, |
| "abs_max": 8.573103904724121, |
| "frac_dead": 7.417467031700653e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.qkv": { |
| "mean": 0.005614753346890211, |
| "std": 0.6067090034484863, |
| "abs_max": 6.34375, |
| "frac_dead": 0.0001014780136756599, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.proj": { |
| "mean": 0.00559053011238575, |
| "std": 0.301848441362381, |
| "abs_max": 1.984375, |
| "frac_dead": 0.0001836352894315496, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.2.op": { |
| "mean": 0.00250375852920115, |
| "std": 0.7678435444831848, |
| "abs_max": 8.5625, |
| "frac_dead": 2.8689702958217822e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm1": { |
| "mean": -0.0012076408602297306, |
| "std": 1.0004695653915405, |
| "abs_max": 10.6675443649292, |
| "frac_dead": 5.695554818885284e-07, |
| "frac_sat": 5.298190686175985e-08 |
| }, |
| "dec_blocks.3.conv1": { |
| "mean": -0.9750791192054749, |
| "std": 2.384197950363159, |
| "abs_max": 42.25, |
| "frac_dead": 1.3007058441871777e-05, |
| "frac_sat": 0.009697622619569302 |
| }, |
| "dec_blocks.3.t_proj": { |
| "mean": -0.000648355926387012, |
| "std": 0.13717307150363922, |
| "abs_max": 0.94140625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm2": { |
| "mean": -0.014358947984874249, |
| "std": 0.9960097074508667, |
| "abs_max": 12.777533531188965, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 2.2596783310291357e-05 |
| }, |
| "dec_blocks.3.conv2": { |
| "mean": 0.018594268709421158, |
| "std": 0.43747326731681824, |
| "abs_max": 6.71875, |
| "frac_dead": 5.581643927143887e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.skip": { |
| "mean": -6.380594277288765e-05, |
| "std": 0.6293055415153503, |
| "abs_max": 6.375, |
| "frac_dead": 7.343292236328125e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.3.op": { |
| "mean": -0.05254688486456871, |
| "std": 0.7604173421859741, |
| "abs_max": 10.375, |
| "frac_dead": 3.1656691135140136e-05, |
| "frac_sat": 1.3245476715439963e-08 |
| }, |
| "dec_blocks.4.norm1": { |
| "mean": -0.0019422242185100913, |
| "std": 0.9994417428970337, |
| "abs_max": 11.856742858886719, |
| "frac_dead": 4.76837158203125e-07, |
| "frac_sat": 2.0751247120642802e-07 |
| }, |
| "dec_blocks.4.conv1": { |
| "mean": -0.7406669855117798, |
| "std": 1.986746072769165, |
| "abs_max": 25.25, |
| "frac_dead": 1.1139445632579736e-05, |
| "frac_sat": 0.0016398695297539234 |
| }, |
| "dec_blocks.4.t_proj": { |
| "mean": -0.004709389992058277, |
| "std": 0.15576112270355225, |
| "abs_max": 0.96484375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm2": { |
| "mean": -0.007681197486817837, |
| "std": 0.9972299933433533, |
| "abs_max": 12.993465423583984, |
| "frac_dead": 1.284811219193216e-06, |
| "frac_sat": 2.211994569734088e-06 |
| }, |
| "dec_blocks.4.conv2": { |
| "mean": -0.046417709439992905, |
| "std": 0.5319167971611023, |
| "abs_max": 5.875, |
| "frac_dead": 6.185637903399765e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.skip": { |
| "mean": -0.02933773770928383, |
| "std": 0.7752041816711426, |
| "abs_max": 7.90625, |
| "frac_dead": 7.099575304891914e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.4.op": { |
| "mean": -0.261995792388916, |
| "std": 1.4503177404403687, |
| "abs_max": 26.75, |
| "frac_dead": 3.314018249511719e-05, |
| "frac_sat": 0.000735077599529177 |
| }, |
| "dec_blocks.5.norm1": { |
| "mean": -0.005145529750734568, |
| "std": 0.997856080532074, |
| "abs_max": 13.626863479614258, |
| "frac_dead": 8.940696716308594e-07, |
| "frac_sat": 1.8460883666193695e-06 |
| }, |
| "dec_blocks.5.conv1": { |
| "mean": -0.1504458636045456, |
| "std": 1.501868724822998, |
| "abs_max": 17.125, |
| "frac_dead": 1.5503830582019873e-05, |
| "frac_sat": 2.1060308426967822e-05 |
| }, |
| "dec_blocks.5.t_proj": { |
| "mean": 0.012189735658466816, |
| "std": 0.24086928367614746, |
| "abs_max": 1.859375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.norm2": { |
| "mean": -0.0016006923979148269, |
| "std": 1.0046899318695068, |
| "abs_max": 12.903286933898926, |
| "frac_dead": 6.324715400296554e-07, |
| "frac_sat": 6.920761848050461e-07 |
| }, |
| "dec_blocks.5.conv2": { |
| "mean": -0.23159591853618622, |
| "std": 1.682431697845459, |
| "abs_max": 21.25, |
| "frac_dead": 2.7279058485873975e-05, |
| "frac_sat": 0.0008494390640407801 |
| }, |
| "dec_blocks.5.skip": { |
| "mean": -0.16567134857177734, |
| "std": 1.1772466897964478, |
| "abs_max": 12.4375, |
| "frac_dead": 6.788306927774101e-05, |
| "frac_sat": 9.238719940185547e-07 |
| }, |
| "out_norm": { |
| "mean": -0.008237053640186787, |
| "std": 0.9936524033546448, |
| "abs_max": 11.3441743850708, |
| "frac_dead": 1.417266048520105e-06, |
| "frac_sat": 3.8743019104003906e-07 |
| }, |
| "out_conv": { |
| "mean": 0.16196849942207336, |
| "std": 0.25798922777175903, |
| "abs_max": 0.9609375, |
| "frac_dead": 1.4834933608653955e-05, |
| "frac_sat": 0.0 |
| } |
| } |
| } |
