| { |
| "iter": 1500, |
| "per_layer": { |
| "time_mlp.1.weight": { |
| "w_norm": 18.831348419189453, |
| "g_norm": 0.10205292701721191, |
| "shape": [ |
| 1024, |
| 256 |
| ] |
| }, |
| "time_mlp.1.bias": { |
| "w_norm": 1.1589568853378296, |
| "g_norm": 0.009027610532939434, |
| "shape": [ |
| 1024 |
| ] |
| }, |
| "time_mlp.3.weight": { |
| "w_norm": 9.640146255493164, |
| "g_norm": 0.019456155598163605, |
| "shape": [ |
| 256, |
| 1024 |
| ] |
| }, |
| "time_mlp.3.bias": { |
| "w_norm": 0.29026785492897034, |
| "g_norm": 0.00706234248355031, |
| "shape": [ |
| 256 |
| ] |
| }, |
| "input_conv.weight": { |
| "w_norm": 7.941542625427246, |
| "g_norm": 0.06458809971809387, |
| "shape": [ |
| 192, |
| 6, |
| 3, |
| 3 |
| ] |
| }, |
| "input_conv.bias": { |
| "w_norm": 1.1898912191390991, |
| "g_norm": 0.07269138097763062, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.weight": { |
| "w_norm": 13.811203002929688, |
| "g_norm": 0.017467472702264786, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.bias": { |
| "w_norm": 0.08192380517721176, |
| "g_norm": 0.026821866631507874, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv1.weight": { |
| "w_norm": 8.32677936553955, |
| "g_norm": 0.17138740420341492, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv1.bias": { |
| "w_norm": 0.20204275846481323, |
| "g_norm": 0.017192741855978966, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.t_proj.weight": { |
| "w_norm": 8.070746421813965, |
| "g_norm": 0.010215070098638535, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.0.t_proj.bias": { |
| "w_norm": 0.5025191903114319, |
| "g_norm": 0.01719454675912857, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.weight": { |
| "w_norm": 13.839881896972656, |
| "g_norm": 0.008044346235692501, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.bias": { |
| "w_norm": 0.04994768276810646, |
| "g_norm": 0.009143427945673466, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv2.weight": { |
| "w_norm": 8.244281768798828, |
| "g_norm": 0.08354029059410095, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv2.bias": { |
| "w_norm": 0.1875186562538147, |
| "g_norm": 0.005785597022622824, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.weight": { |
| "w_norm": 13.85268497467041, |
| "g_norm": 7.552062015747651e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.bias": { |
| "w_norm": 0.026036987081170082, |
| "g_norm": 4.653045107261278e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv1.weight": { |
| "w_norm": 8.178509712219238, |
| "g_norm": 0.001746668480336666, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv1.bias": { |
| "w_norm": 0.19200706481933594, |
| "g_norm": 5.417833017418161e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.t_proj.weight": { |
| "w_norm": 8.048393249511719, |
| "g_norm": 8.108843758236617e-05, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.1.t_proj.bias": { |
| "w_norm": 0.5217288136482239, |
| "g_norm": 5.417064676294103e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.weight": { |
| "w_norm": 13.822420120239258, |
| "g_norm": 0.00010076359467348084, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.bias": { |
| "w_norm": 0.04598449543118477, |
| "g_norm": 6.376473902491853e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv2.weight": { |
| "w_norm": 8.135522842407227, |
| "g_norm": 0.002183968434110284, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv2.bias": { |
| "w_norm": 0.20787833631038666, |
| "g_norm": 0.00010054003359982744, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.weight": { |
| "w_norm": 13.844120025634766, |
| "g_norm": 2.4686974938958883e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.bias": { |
| "w_norm": 0.035012636333703995, |
| "g_norm": 1.9267890820628963e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.conv1.weight": { |
| "w_norm": 11.575699806213379, |
| "g_norm": 0.000611058552749455, |
| "shape": [ |
| 384, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv1.bias": { |
| "w_norm": 0.26600784063339233, |
| "g_norm": 2.2671565602649935e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.t_proj.weight": { |
| "w_norm": 11.522205352783203, |
| "g_norm": 3.756795194931328e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.2.t_proj.bias": { |
| "w_norm": 0.720262348651886, |
| "g_norm": 2.2679385438095778e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.weight": { |
| "w_norm": 19.48476791381836, |
| "g_norm": 4.690673085860908e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.bias": { |
| "w_norm": 0.11410542577505112, |
| "g_norm": 2.5380313672940247e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.conv2.weight": { |
| "w_norm": 11.546863555908203, |
| "g_norm": 0.0008082764106802642, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv2.bias": { |
| "w_norm": 0.20199169218540192, |
| "g_norm": 2.2719479602528736e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.skip.weight": { |
| "w_norm": 11.241255760192871, |
| "g_norm": 0.0006491548265330493, |
| "shape": [ |
| 384, |
| 192, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.2.skip.bias": { |
| "w_norm": 0.8258646726608276, |
| "g_norm": 2.2719479602528736e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.weight": { |
| "w_norm": 19.59480857849121, |
| "g_norm": 4.11441669712076e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.bias": { |
| "w_norm": 0.03167359158396721, |
| "g_norm": 3.48591424881306e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv1.weight": { |
| "w_norm": 11.656805038452148, |
| "g_norm": 0.00011399946379242465, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv1.bias": { |
| "w_norm": 0.1977333277463913, |
| "g_norm": 3.2061070669442415e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.t_proj.weight": { |
| "w_norm": 11.358353614807129, |
| "g_norm": 2.2405642994272057e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.3.t_proj.bias": { |
| "w_norm": 0.6917930841445923, |
| "g_norm": 3.2053535505838227e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.weight": { |
| "w_norm": 19.485275268554688, |
| "g_norm": 5.654585947922897e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.bias": { |
| "w_norm": 0.0983862653374672, |
| "g_norm": 4.409213943290524e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv2.weight": { |
| "w_norm": 11.416952133178711, |
| "g_norm": 0.00018496303528081626, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv2.bias": { |
| "w_norm": 0.1904715597629547, |
| "g_norm": 6.365587978507392e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.weight": { |
| "w_norm": 19.591276168823242, |
| "g_norm": 1.3642720659845509e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.bias": { |
| "w_norm": 0.03558076173067093, |
| "g_norm": 1.0926190725513152e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.conv1.weight": { |
| "w_norm": 16.351974487304688, |
| "g_norm": 4.5458058593794703e-05, |
| "shape": [ |
| 768, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv1.bias": { |
| "w_norm": 0.2806565761566162, |
| "g_norm": 7.310076739486249e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.t_proj.weight": { |
| "w_norm": 16.041704177856445, |
| "g_norm": 8.280318866127345e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.4.t_proj.bias": { |
| "w_norm": 0.9907169342041016, |
| "g_norm": 7.310062528631533e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.weight": { |
| "w_norm": 27.56458282470703, |
| "g_norm": 1.4526040104101412e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.bias": { |
| "w_norm": 0.1475459188222885, |
| "g_norm": 1.0344501788495108e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.conv2.weight": { |
| "w_norm": 16.167402267456055, |
| "g_norm": 7.684766023885459e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv2.bias": { |
| "w_norm": 0.19410400092601776, |
| "g_norm": 2.097709511872381e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.skip.weight": { |
| "w_norm": 15.907557487487793, |
| "g_norm": 0.000103915692307055, |
| "shape": [ |
| 768, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.4.skip.bias": { |
| "w_norm": 0.8080542683601379, |
| "g_norm": 2.097709511872381e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.weight": { |
| "w_norm": 27.710412979125977, |
| "g_norm": 5.43541773367906e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.bias": { |
| "w_norm": 0.049210891127586365, |
| "g_norm": 5.415469104264048e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv1.weight": { |
| "w_norm": 16.535066604614258, |
| "g_norm": 1.713006349746138e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv1.bias": { |
| "w_norm": 0.20113901793956757, |
| "g_norm": 3.853272971809929e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.t_proj.weight": { |
| "w_norm": 15.998994827270508, |
| "g_norm": 3.5397795272729127e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.5.t_proj.bias": { |
| "w_norm": 0.9972907304763794, |
| "g_norm": 3.853358805372409e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.weight": { |
| "w_norm": 27.60606575012207, |
| "g_norm": 5.676409955412964e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.bias": { |
| "w_norm": 0.1123356744647026, |
| "g_norm": 5.299498866406793e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv2.weight": { |
| "w_norm": 16.294984817504883, |
| "g_norm": 3.0194179998943582e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv2.bias": { |
| "w_norm": 0.19272859394550323, |
| "g_norm": 7.503235224248783e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.3.norm.weight": { |
| "w_norm": 19.56330680847168, |
| "g_norm": 4.690531568485312e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.norm.bias": { |
| "w_norm": 0.04402874410152435, |
| "g_norm": 2.7959677026956342e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.qkv.weight": { |
| "w_norm": 19.674089431762695, |
| "g_norm": 0.0001051163490046747, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.qkv.bias": { |
| "w_norm": 1.0076570510864258, |
| "g_norm": 4.560146862786496e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "enc_attns.3.proj.weight": { |
| "w_norm": 11.307135581970215, |
| "g_norm": 8.055998478084803e-05, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.proj.bias": { |
| "w_norm": 0.5636854767799377, |
| "g_norm": 4.890626769338269e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.4.norm.weight": { |
| "w_norm": 27.659299850463867, |
| "g_norm": 1.196475409415143e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.norm.bias": { |
| "w_norm": 0.03487004339694977, |
| "g_norm": 7.422240173582395e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.qkv.weight": { |
| "w_norm": 27.882322311401367, |
| "g_norm": 4.89087360620033e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.qkv.bias": { |
| "w_norm": 1.0131044387817383, |
| "g_norm": 1.190264356409898e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.4.proj.weight": { |
| "w_norm": 15.986047744750977, |
| "g_norm": 4.969876681570895e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.proj.bias": { |
| "w_norm": 0.5729395747184753, |
| "g_norm": 1.8785445945468382e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.weight": { |
| "w_norm": 27.67577362060547, |
| "g_norm": 4.395630526232708e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.bias": { |
| "w_norm": 0.035282380878925323, |
| "g_norm": 2.870086461825849e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.qkv.weight": { |
| "w_norm": 27.845542907714844, |
| "g_norm": 1.6994632460409775e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.qkv.bias": { |
| "w_norm": 0.9989751577377319, |
| "g_norm": 4.274110381174978e-07, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.5.proj.weight": { |
| "w_norm": 16.010074615478516, |
| "g_norm": 1.648172110435553e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.proj.bias": { |
| "w_norm": 0.5966185331344604, |
| "g_norm": 7.092186251611565e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_downs.0.op.weight": { |
| "w_norm": 8.130995750427246, |
| "g_norm": 0.0033071700017899275, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.0.op.bias": { |
| "w_norm": 0.18532414734363556, |
| "g_norm": 0.00010940373613266274, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.1.op.weight": { |
| "w_norm": 8.075477600097656, |
| "g_norm": 0.0010724548483267426, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.1.op.bias": { |
| "w_norm": 0.19328613579273224, |
| "g_norm": 2.2360094590112567e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.2.op.weight": { |
| "w_norm": 11.434945106506348, |
| "g_norm": 0.000316201476380229, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.2.op.bias": { |
| "w_norm": 0.18795061111450195, |
| "g_norm": 7.141250534914434e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.3.op.weight": { |
| "w_norm": 11.38402271270752, |
| "g_norm": 0.00017456771456636488, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.3.op.bias": { |
| "w_norm": 0.1932166963815689, |
| "g_norm": 1.5802761481609195e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.4.op.weight": { |
| "w_norm": 16.206579208374023, |
| "g_norm": 9.180299821309745e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.4.op.bias": { |
| "w_norm": 0.19638435542583466, |
| "g_norm": 7.905038046374102e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.weight": { |
| "w_norm": 27.71148681640625, |
| "g_norm": 2.901707603086834e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.bias": { |
| "w_norm": 0.04520085081458092, |
| "g_norm": 3.0601083267356444e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv1.weight": { |
| "w_norm": 16.51447296142578, |
| "g_norm": 8.964741937234066e-06, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv1.bias": { |
| "w_norm": 0.20231729745864868, |
| "g_norm": 2.1629281832247216e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.t_proj.weight": { |
| "w_norm": 16.008007049560547, |
| "g_norm": 2.066750681706253e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid1.t_proj.bias": { |
| "w_norm": 1.0318657159805298, |
| "g_norm": 2.1627558055570262e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.weight": { |
| "w_norm": 27.610170364379883, |
| "g_norm": 2.3066324672527116e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.bias": { |
| "w_norm": 0.12007729709148407, |
| "g_norm": 3.1912145459500607e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv2.weight": { |
| "w_norm": 16.278959274291992, |
| "g_norm": 1.5227375115500763e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv2.bias": { |
| "w_norm": 0.19529534876346588, |
| "g_norm": 4.908901587441505e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.weight": { |
| "w_norm": 27.68242835998535, |
| "g_norm": 1.5732706515336758e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.bias": { |
| "w_norm": 0.03362787514925003, |
| "g_norm": 1.7879504810025537e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.qkv.weight": { |
| "w_norm": 27.83576011657715, |
| "g_norm": 6.8316489887365606e-06, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.qkv.bias": { |
| "w_norm": 0.9854804277420044, |
| "g_norm": 2.807340706567629e-07, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "mid_attn.proj.weight": { |
| "w_norm": 16.009660720825195, |
| "g_norm": 6.5837921283673495e-06, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.proj.bias": { |
| "w_norm": 0.5765897631645203, |
| "g_norm": 4.809566576113866e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.weight": { |
| "w_norm": 27.716794967651367, |
| "g_norm": 1.7296281384915346e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.bias": { |
| "w_norm": 0.05131439492106438, |
| "g_norm": 1.5387250584808498e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv1.weight": { |
| "w_norm": 16.50710105895996, |
| "g_norm": 7.962901690916624e-06, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv1.bias": { |
| "w_norm": 0.19801872968673706, |
| "g_norm": 1.5266020625404053e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.t_proj.weight": { |
| "w_norm": 15.995810508728027, |
| "g_norm": 2.089972923613459e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid2.t_proj.bias": { |
| "w_norm": 0.9966025352478027, |
| "g_norm": 1.5270575204340275e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.weight": { |
| "w_norm": 27.59794807434082, |
| "g_norm": 1.9348948399056098e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.bias": { |
| "w_norm": 0.1285128891468048, |
| "g_norm": 2.489040582531743e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv2.weight": { |
| "w_norm": 16.23427963256836, |
| "g_norm": 1.610977233212907e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv2.bias": { |
| "w_norm": 0.18982110917568207, |
| "g_norm": 4.8388949380751e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm1.weight": { |
| "w_norm": 39.197998046875, |
| "g_norm": 3.1209330586534634e-07, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.norm1.bias": { |
| "w_norm": 0.06604019552469254, |
| "g_norm": 2.441701383304462e-07, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.conv1.weight": { |
| "w_norm": 17.042268753051758, |
| "g_norm": 1.324670301983133e-05, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv1.bias": { |
| "w_norm": 0.14318080246448517, |
| "g_norm": 1.5126960306588444e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.t_proj.weight": { |
| "w_norm": 16.01215171813965, |
| "g_norm": 1.9884710411588458e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.0.t_proj.bias": { |
| "w_norm": 1.0212759971618652, |
| "g_norm": 1.5122640206755023e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.weight": { |
| "w_norm": 27.559242248535156, |
| "g_norm": 3.4676895666052587e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.bias": { |
| "w_norm": 0.16142374277114868, |
| "g_norm": 4.399703641411179e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.conv2.weight": { |
| "w_norm": 16.135143280029297, |
| "g_norm": 3.1659994419896975e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv2.bias": { |
| "w_norm": 0.19768120348453522, |
| "g_norm": 1.10852681700635e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.skip.weight": { |
| "w_norm": 15.865204811096191, |
| "g_norm": 0.00012172277638455853, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.0.skip.bias": { |
| "w_norm": 0.4107856750488281, |
| "g_norm": 1.10852681700635e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm1.weight": { |
| "w_norm": 39.19361877441406, |
| "g_norm": 1.6795871715657995e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.norm1.bias": { |
| "w_norm": 0.05957057327032089, |
| "g_norm": 1.083103825294529e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.conv1.weight": { |
| "w_norm": 17.069610595703125, |
| "g_norm": 4.9241381930187345e-05, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv1.bias": { |
| "w_norm": 0.1401885598897934, |
| "g_norm": 5.735878403356764e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.t_proj.weight": { |
| "w_norm": 16.04375457763672, |
| "g_norm": 8.650154086353723e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.1.t_proj.bias": { |
| "w_norm": 0.9863389730453491, |
| "g_norm": 5.737327342103526e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.weight": { |
| "w_norm": 27.54302215576172, |
| "g_norm": 1.7761914250513655e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.bias": { |
| "w_norm": 0.19749881327152252, |
| "g_norm": 2.196136847487651e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.conv2.weight": { |
| "w_norm": 16.147022247314453, |
| "g_norm": 0.00012046505435137078, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv2.bias": { |
| "w_norm": 0.19546672701835632, |
| "g_norm": 3.8538873923243955e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.skip.weight": { |
| "w_norm": 15.8510160446167, |
| "g_norm": 0.00025108447880484164, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.1.skip.bias": { |
| "w_norm": 0.4066181778907776, |
| "g_norm": 3.8538873923243955e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.2.norm1.weight": { |
| "w_norm": 33.947349548339844, |
| "g_norm": 2.5137094326055376e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.norm1.bias": { |
| "w_norm": 0.06189341843128204, |
| "g_norm": 2.2232970877666958e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.conv1.weight": { |
| "w_norm": 12.084885597229004, |
| "g_norm": 0.00012408540351316333, |
| "shape": [ |
| 384, |
| 1152, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv1.bias": { |
| "w_norm": 0.1179933249950409, |
| "g_norm": 2.0747602320625447e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.t_proj.weight": { |
| "w_norm": 11.32999038696289, |
| "g_norm": 3.524335625115782e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.2.t_proj.bias": { |
| "w_norm": 0.7094293236732483, |
| "g_norm": 2.0747856979141943e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.weight": { |
| "w_norm": 19.48737335205078, |
| "g_norm": 3.282286343164742e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.bias": { |
| "w_norm": 0.131941020488739, |
| "g_norm": 5.007308573112823e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.conv2.weight": { |
| "w_norm": 11.395310401916504, |
| "g_norm": 0.00022668339079245925, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv2.bias": { |
| "w_norm": 0.18954412639141083, |
| "g_norm": 1.244637678610161e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.skip.weight": { |
| "w_norm": 11.1150541305542, |
| "g_norm": 0.0005784353706985712, |
| "shape": [ |
| 384, |
| 1152, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.2.skip.bias": { |
| "w_norm": 0.3310128450393677, |
| "g_norm": 1.244637678610161e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm1.weight": { |
| "w_norm": 27.713695526123047, |
| "g_norm": 1.6070203855633736e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.norm1.bias": { |
| "w_norm": 0.06734566390514374, |
| "g_norm": 1.080029960576212e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.conv1.weight": { |
| "w_norm": 11.990684509277344, |
| "g_norm": 0.0005816977354697883, |
| "shape": [ |
| 384, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv1.bias": { |
| "w_norm": 0.14317339658737183, |
| "g_norm": 9.035586117533967e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.t_proj.weight": { |
| "w_norm": 11.34940242767334, |
| "g_norm": 2.0452938770176843e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.3.t_proj.bias": { |
| "w_norm": 0.7089352011680603, |
| "g_norm": 9.036564733833075e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.weight": { |
| "w_norm": 19.466581344604492, |
| "g_norm": 1.5670944776502438e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.bias": { |
| "w_norm": 0.1333981603384018, |
| "g_norm": 1.8773333067656495e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.conv2.weight": { |
| "w_norm": 11.425914764404297, |
| "g_norm": 0.0009791628690436482, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv2.bias": { |
| "w_norm": 0.19313561916351318, |
| "g_norm": 4.651550989365205e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.skip.weight": { |
| "w_norm": 11.149859428405762, |
| "g_norm": 0.0015515109989792109, |
| "shape": [ |
| 384, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.3.skip.bias": { |
| "w_norm": 0.4057067632675171, |
| "g_norm": 4.651550989365205e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.4.norm1.weight": { |
| "w_norm": 23.98821258544922, |
| "g_norm": 5.4541109420824796e-05, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.norm1.bias": { |
| "w_norm": 0.06465596705675125, |
| "g_norm": 4.13734233006835e-05, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.conv1.weight": { |
| "w_norm": 8.374838829040527, |
| "g_norm": 0.001791031681932509, |
| "shape": [ |
| 192, |
| 576, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv1.bias": { |
| "w_norm": 0.11661636829376221, |
| "g_norm": 4.558874206850305e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.t_proj.weight": { |
| "w_norm": 8.018696784973145, |
| "g_norm": 6.551636761287227e-05, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.4.t_proj.bias": { |
| "w_norm": 0.4810430109500885, |
| "g_norm": 4.55839435744565e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.weight": { |
| "w_norm": 13.801824569702148, |
| "g_norm": 9.073015098692849e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.bias": { |
| "w_norm": 0.05389580875635147, |
| "g_norm": 0.00011636758426902816, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.conv2.weight": { |
| "w_norm": 8.050837516784668, |
| "g_norm": 0.004189273342490196, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv2.bias": { |
| "w_norm": 0.19405966997146606, |
| "g_norm": 0.0002257868618471548, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.skip.weight": { |
| "w_norm": 7.899153709411621, |
| "g_norm": 0.004444679711014032, |
| "shape": [ |
| 192, |
| 576, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.4.skip.bias": { |
| "w_norm": 0.34669438004493713, |
| "g_norm": 0.0002257868618471548, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm1.weight": { |
| "w_norm": 19.55978775024414, |
| "g_norm": 0.0025432000402361155, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.norm1.bias": { |
| "w_norm": 0.08432969450950623, |
| "g_norm": 0.0032189551275223494, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.conv1.weight": { |
| "w_norm": 8.425467491149902, |
| "g_norm": 0.04127212241292, |
| "shape": [ |
| 192, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv1.bias": { |
| "w_norm": 0.13819772005081177, |
| "g_norm": 0.0018730823649093509, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.t_proj.weight": { |
| "w_norm": 8.143948554992676, |
| "g_norm": 0.0016058351611718535, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.5.t_proj.bias": { |
| "w_norm": 0.504915714263916, |
| "g_norm": 0.0018739085644483566, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.weight": { |
| "w_norm": 13.880186080932617, |
| "g_norm": 0.002454320201650262, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.bias": { |
| "w_norm": 0.03555542230606079, |
| "g_norm": 0.002779952948912978, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.conv2.weight": { |
| "w_norm": 8.265290260314941, |
| "g_norm": 0.053190335631370544, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv2.bias": { |
| "w_norm": 0.1995941400527954, |
| "g_norm": 0.002878693165257573, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.skip.weight": { |
| "w_norm": 8.083870887756348, |
| "g_norm": 0.050536688417196274, |
| "shape": [ |
| 192, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.5.skip.bias": { |
| "w_norm": 0.42189013957977295, |
| "g_norm": 0.002878693165257573, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_attns.0.norm.weight": { |
| "w_norm": 27.65727424621582, |
| "g_norm": 6.846094038337469e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.norm.bias": { |
| "w_norm": 0.03050125576555729, |
| "g_norm": 4.60043281691469e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.qkv.weight": { |
| "w_norm": 27.862565994262695, |
| "g_norm": 2.4649789338582195e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.qkv.bias": { |
| "w_norm": 1.0089398622512817, |
| "g_norm": 6.871725872770185e-07, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.0.proj.weight": { |
| "w_norm": 16.023706436157227, |
| "g_norm": 2.3888020223239437e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.proj.bias": { |
| "w_norm": 0.5768023133277893, |
| "g_norm": 1.100436406886729e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.weight": { |
| "w_norm": 27.667020797729492, |
| "g_norm": 1.563885575706081e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.bias": { |
| "w_norm": 0.033365633338689804, |
| "g_norm": 1.447783574803907e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.qkv.weight": { |
| "w_norm": 27.891311645507812, |
| "g_norm": 6.250386650208384e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.qkv.bias": { |
| "w_norm": 1.0041871070861816, |
| "g_norm": 2.3000570763542783e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.1.proj.weight": { |
| "w_norm": 16.047632217407227, |
| "g_norm": 5.869360757060349e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.proj.bias": { |
| "w_norm": 0.579659640789032, |
| "g_norm": 3.644436446847976e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.2.norm.weight": { |
| "w_norm": 19.534183502197266, |
| "g_norm": 4.331167474447284e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.norm.bias": { |
| "w_norm": 0.019406652078032494, |
| "g_norm": 4.485439148993464e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.qkv.weight": { |
| "w_norm": 19.665908813476562, |
| "g_norm": 0.0001103595714084804, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.qkv.bias": { |
| "w_norm": 0.9858006238937378, |
| "g_norm": 7.38036851544166e-06, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_attns.2.proj.weight": { |
| "w_norm": 11.296539306640625, |
| "g_norm": 0.00010973431199090555, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.proj.bias": { |
| "w_norm": 0.5865283012390137, |
| "g_norm": 1.2120593964937143e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.0.op.weight": { |
| "w_norm": 16.096851348876953, |
| "g_norm": 0.00019765795150306076, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.0.op.bias": { |
| "w_norm": 0.2011629343032837, |
| "g_norm": 1.6096408899102244e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.1.op.weight": { |
| "w_norm": 16.08983039855957, |
| "g_norm": 0.0004939130740240216, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.1.op.bias": { |
| "w_norm": 0.18965400755405426, |
| "g_norm": 5.676467935700202e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.2.op.weight": { |
| "w_norm": 11.300466537475586, |
| "g_norm": 0.0012292304309085011, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.2.op.bias": { |
| "w_norm": 0.1965014785528183, |
| "g_norm": 2.0350802515167743e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.3.op.weight": { |
| "w_norm": 11.314447402954102, |
| "g_norm": 0.0029695341363549232, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.3.op.bias": { |
| "w_norm": 0.19801823794841766, |
| "g_norm": 8.36284234537743e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.4.op.weight": { |
| "w_norm": 7.997620105743408, |
| "g_norm": 0.014928634278476238, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.4.op.bias": { |
| "w_norm": 0.19552351534366608, |
| "g_norm": 0.00039662071503698826, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.weight": { |
| "w_norm": 13.799749374389648, |
| "g_norm": 0.0063019501976668835, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.bias": { |
| "w_norm": 0.07100910693407059, |
| "g_norm": 0.009717518463730812, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_conv.weight": { |
| "w_norm": 0.12911434471607208, |
| "g_norm": 0.9639920592308044, |
| "shape": [ |
| 3, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "out_conv.bias": { |
| "w_norm": 0.00815030001103878, |
| "g_norm": 0.06712251156568527, |
| "shape": [ |
| 3 |
| ] |
| } |
| }, |
| "activations": { |
| "input_conv": { |
| "mean": 0.00026234411052428186, |
| "std": 0.3449311852455139, |
| "abs_max": 2.703125, |
| "frac_dead": 0.000533885438926518, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.1": { |
| "mean": 0.021856388077139854, |
| "std": 0.2685975730419159, |
| "abs_max": 3.671875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.3": { |
| "mean": 0.02677418850362301, |
| "std": 0.18945495784282684, |
| "abs_max": 0.78515625, |
| "frac_dead": 0.00032552084303461015, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm1": { |
| "mean": -0.004714304581284523, |
| "std": 0.9962726831436157, |
| "abs_max": 6.979825019836426, |
| "frac_dead": 7.980399914231384e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv1": { |
| "mean": -0.027971331030130386, |
| "std": 0.5426817536354065, |
| "abs_max": 9.0625, |
| "frac_dead": 6.779366231057793e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.t_proj": { |
| "mean": -0.0008105086162686348, |
| "std": 0.17592351138591766, |
| "abs_max": 1.4296875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm2": { |
| "mean": -0.002945144195109606, |
| "std": 0.9993420839309692, |
| "abs_max": 10.731555938720703, |
| "frac_dead": 6.622738624173508e-07, |
| "frac_sat": 2.9802322387695312e-08 |
| }, |
| "enc_blocks.0.conv2": { |
| "mean": -0.010785315185785294, |
| "std": 0.8624541163444519, |
| "abs_max": 8.125, |
| "frac_dead": 3.263685357524082e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.0.op": { |
| "mean": -0.10003193467855453, |
| "std": 1.335770606994629, |
| "abs_max": 12.875, |
| "frac_dead": 1.9497341781971045e-05, |
| "frac_sat": 2.384185791015625e-07 |
| }, |
| "enc_blocks.1.norm1": { |
| "mean": -0.0012766718864440918, |
| "std": 0.9998140931129456, |
| "abs_max": 9.364228248596191, |
| "frac_dead": 4.106097719613899e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv1": { |
| "mean": -0.24094407260417938, |
| "std": 1.0454233884811401, |
| "abs_max": 10.25, |
| "frac_dead": 3.066327917622402e-05, |
| "frac_sat": 1.3245476715439963e-08 |
| }, |
| "enc_blocks.1.t_proj": { |
| "mean": 0.00418504374101758, |
| "std": 0.1214127168059349, |
| "abs_max": 0.86328125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.norm2": { |
| "mean": -0.003717670915648341, |
| "std": 0.9974992871284485, |
| "abs_max": 8.737704277038574, |
| "frac_dead": 3.97364289028701e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv2": { |
| "mean": 0.028257178142666817, |
| "std": 0.7950882911682129, |
| "abs_max": 9.4375, |
| "frac_dead": 5.786948895547539e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.1.op": { |
| "mean": -0.054014790803194046, |
| "std": 1.8062548637390137, |
| "abs_max": 11.375, |
| "frac_dead": 1.9550323486328125e-05, |
| "frac_sat": 7.417467031700653e-07 |
| }, |
| "enc_blocks.2.norm1": { |
| "mean": -0.001649506390094757, |
| "std": 0.9994705319404602, |
| "abs_max": 7.457204818725586, |
| "frac_dead": 1.0066562481370056e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.conv1": { |
| "mean": -0.19917495548725128, |
| "std": 1.0274417400360107, |
| "abs_max": 11.5, |
| "frac_dead": 4.151132452534512e-05, |
| "frac_sat": 1.5099843722055084e-06 |
| }, |
| "enc_blocks.2.t_proj": { |
| "mean": -0.002049730159342289, |
| "std": 0.17591041326522827, |
| "abs_max": 1.4296875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.norm2": { |
| "mean": -0.006312957499176264, |
| "std": 0.9954407811164856, |
| "abs_max": 13.2487211227417, |
| "frac_dead": 3.178914482759865e-07, |
| "frac_sat": 1.8543667010817444e-06 |
| }, |
| "enc_blocks.2.conv2": { |
| "mean": 0.06416342407464981, |
| "std": 0.6680892705917358, |
| "abs_max": 6.40625, |
| "frac_dead": 3.208054476999678e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.skip": { |
| "mean": -0.02201053313910961, |
| "std": 1.0272172689437866, |
| "abs_max": 8.375, |
| "frac_dead": 0.00010784467303892598, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.2.op": { |
| "mean": 0.13211365044116974, |
| "std": 1.7092348337173462, |
| "abs_max": 9.9375, |
| "frac_dead": 1.33514404296875e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.norm1": { |
| "mean": 6.256169581320137e-05, |
| "std": 1.0005357265472412, |
| "abs_max": 8.886502265930176, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.conv1": { |
| "mean": -0.1537984013557434, |
| "std": 1.3866938352584839, |
| "abs_max": 20.75, |
| "frac_dead": 2.1404690414783545e-05, |
| "frac_sat": 0.0002346038818359375 |
| }, |
| "enc_blocks.3.t_proj": { |
| "mean": 0.0013901263009756804, |
| "std": 0.09665051102638245, |
| "abs_max": 0.56640625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.norm2": { |
| "mean": -0.005035887472331524, |
| "std": 0.9948577284812927, |
| "abs_max": 11.17821979522705, |
| "frac_dead": 5.298190899338806e-07, |
| "frac_sat": 2.013312496274011e-06 |
| }, |
| "enc_blocks.3.conv2": { |
| "mean": -0.0170371625572443, |
| "std": 0.7531740069389343, |
| "abs_max": 10.875, |
| "frac_dead": 3.4332275390625e-05, |
| "frac_sat": 1.059638137235197e-07 |
| }, |
| "enc_attns.3.norm": { |
| "mean": 0.00035631077480502427, |
| "std": 0.9987042546272278, |
| "abs_max": 11.254258155822754, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 3.178914482759865e-07 |
| }, |
| "enc_attns.3.qkv": { |
| "mean": -0.0017836717888712883, |
| "std": 0.6752532720565796, |
| "abs_max": 10.125, |
| "frac_dead": 9.808717004489154e-05, |
| "frac_sat": 3.5321271241173235e-08 |
| }, |
| "enc_attns.3.proj": { |
| "mean": -0.01955404318869114, |
| "std": 0.49900659918785095, |
| "abs_max": 5.0625, |
| "frac_dead": 0.00014368693518918008, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.3.op": { |
| "mean": -0.12393307685852051, |
| "std": 2.2092812061309814, |
| "abs_max": 13.125, |
| "frac_dead": 1.6530355424038135e-05, |
| "frac_sat": 0.00044462416553869843 |
| }, |
| "enc_blocks.4.norm1": { |
| "mean": -0.00021128480148036033, |
| "std": 1.0009337663650513, |
| "abs_max": 10.801759719848633, |
| "frac_dead": 2.1192763597355224e-06, |
| "frac_sat": 1.6954210195763153e-06 |
| }, |
| "enc_blocks.4.conv1": { |
| "mean": -0.17340169847011566, |
| "std": 1.5200656652450562, |
| "abs_max": 26.625, |
| "frac_dead": 2.288818359375e-05, |
| "frac_sat": 0.000911712646484375 |
| }, |
| "enc_blocks.4.t_proj": { |
| "mean": -0.0030492052901536226, |
| "std": 0.08274214714765549, |
| "abs_max": 0.5625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.norm2": { |
| "mean": -0.005301392171531916, |
| "std": 0.9954476952552795, |
| "abs_max": 16.856233596801758, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 5.91278076171875e-05 |
| }, |
| "enc_blocks.4.conv2": { |
| "mean": 0.03106667287647724, |
| "std": 0.6966545581817627, |
| "abs_max": 5.6875, |
| "frac_dead": 2.47955322265625e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.skip": { |
| "mean": -0.03207727149128914, |
| "std": 1.269452691078186, |
| "abs_max": 9.3125, |
| "frac_dead": 5.955166489002295e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.norm": { |
| "mean": 2.906637018895708e-05, |
| "std": 0.9980901479721069, |
| "abs_max": 10.796359062194824, |
| "frac_dead": 1.0596381798677612e-06, |
| "frac_sat": 2.119276274470394e-07 |
| }, |
| "enc_attns.4.qkv": { |
| "mean": -0.006540832109749317, |
| "std": 0.6688223481178284, |
| "abs_max": 7.96875, |
| "frac_dead": 7.191410986706614e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.proj": { |
| "mean": -0.0005021503893658519, |
| "std": 0.3761708736419678, |
| "abs_max": 3.6875, |
| "frac_dead": 0.00011041429388569668, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.4.op": { |
| "mean": 0.05056021735072136, |
| "std": 2.155576705932617, |
| "abs_max": 14.1875, |
| "frac_dead": 8.47710543894209e-06, |
| "frac_sat": 0.0006747775478288531 |
| }, |
| "enc_blocks.5.norm1": { |
| "mean": -5.367605263018049e-05, |
| "std": 1.00002121925354, |
| "abs_max": 8.307374000549316, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv1": { |
| "mean": -0.35223713517189026, |
| "std": 1.462549090385437, |
| "abs_max": 19.5, |
| "frac_dead": 2.1192763597355224e-05, |
| "frac_sat": 0.0005204942426644266 |
| }, |
| "enc_blocks.5.t_proj": { |
| "mean": -0.0026074519846588373, |
| "std": 0.07342220097780228, |
| "abs_max": 0.46484375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.norm2": { |
| "mean": -0.004012688063085079, |
| "std": 0.9965819716453552, |
| "abs_max": 10.07455825805664, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 8.477105097881577e-07 |
| }, |
| "enc_blocks.5.conv2": { |
| "mean": 0.08693259209394455, |
| "std": 1.073614478111267, |
| "abs_max": 6.3125, |
| "frac_dead": 2.0345052689663135e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.norm": { |
| "mean": 9.941578900907189e-05, |
| "std": 0.998691201210022, |
| "abs_max": 7.955250263214111, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.qkv": { |
| "mean": -0.008100636303424835, |
| "std": 0.7708551287651062, |
| "abs_max": 5.625, |
| "frac_dead": 7.572880713269114e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.proj": { |
| "mean": -0.008143105544149876, |
| "std": 0.2769840359687805, |
| "abs_max": 3.234375, |
| "frac_dead": 0.00019073486328125, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm1": { |
| "mean": 0.00015265148249454796, |
| "std": 1.000354528427124, |
| "abs_max": 7.3556437492370605, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv1": { |
| "mean": -0.569250762462616, |
| "std": 1.7395997047424316, |
| "abs_max": 22.875, |
| "frac_dead": 2.5431314497836865e-05, |
| "frac_sat": 0.0027398003730922937 |
| }, |
| "mid1.t_proj": { |
| "mean": 0.0030422655399888754, |
| "std": 0.07387731224298477, |
| "abs_max": 0.361328125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm2": { |
| "mean": -0.0043882750906050205, |
| "std": 0.9972242712974548, |
| "abs_max": 10.796936988830566, |
| "frac_dead": 2.543131586207892e-06, |
| "frac_sat": 3.3908420391526306e-06 |
| }, |
| "mid1.conv2": { |
| "mean": 0.01897788606584072, |
| "std": 0.6111142635345459, |
| "abs_max": 4.125, |
| "frac_dead": 2.458360540913418e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.norm": { |
| "mean": 4.941597580909729e-05, |
| "std": 0.9988196492195129, |
| "abs_max": 7.046948432922363, |
| "frac_dead": 2.543131586207892e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.qkv": { |
| "mean": 0.005729823373258114, |
| "std": 0.7872936725616455, |
| "abs_max": 6.1875, |
| "frac_dead": 6.640399078605697e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.proj": { |
| "mean": -0.005338292568922043, |
| "std": 0.24077516794204712, |
| "abs_max": 1.765625, |
| "frac_dead": 0.00017547607421875, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm1": { |
| "mean": 0.00037162296939641237, |
| "std": 1.0006673336029053, |
| "abs_max": 7.135955333709717, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv1": { |
| "mean": -0.6954900622367859, |
| "std": 1.9247785806655884, |
| "abs_max": 24.75, |
| "frac_dead": 2.1192763597355224e-05, |
| "frac_sat": 0.005978054367005825 |
| }, |
| "mid2.t_proj": { |
| "mean": 0.0007212297641672194, |
| "std": 0.07543076574802399, |
| "abs_max": 0.4609375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm2": { |
| "mean": -0.005086460150778294, |
| "std": 0.9972400665283203, |
| "abs_max": 9.937588691711426, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv2": { |
| "mean": 0.004386311862617731, |
| "std": 0.6135098338127136, |
| "abs_max": 4.0625, |
| "frac_dead": 2.7126736313221045e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm1": { |
| "mean": 0.00032785534858703613, |
| "std": 1.0008313655853271, |
| "abs_max": 8.297355651855469, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.conv1": { |
| "mean": -2.0214412212371826, |
| "std": 3.9078080654144287, |
| "abs_max": 50.0, |
| "frac_dead": 3.3908420391526306e-06, |
| "frac_sat": 0.05213843658566475 |
| }, |
| "dec_blocks.0.t_proj": { |
| "mean": -0.001085665193386376, |
| "std": 0.077158622443676, |
| "abs_max": 0.361328125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm2": { |
| "mean": -0.006708974484354258, |
| "std": 0.9966805577278137, |
| "abs_max": 10.097886085510254, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 1.6954210195763153e-06 |
| }, |
| "dec_blocks.0.conv2": { |
| "mean": 0.011965113691985607, |
| "std": 0.45058712363243103, |
| "abs_max": 2.140625, |
| "frac_dead": 2.7126736313221045e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.skip": { |
| "mean": -0.00819339882582426, |
| "std": 1.3349742889404297, |
| "abs_max": 9.375, |
| "frac_dead": 2.8822158128605224e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.norm": { |
| "mean": -7.761849701637402e-06, |
| "std": 0.9980735778808594, |
| "abs_max": 6.372596740722656, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.qkv": { |
| "mean": -0.0017916549695655704, |
| "std": 0.6728198528289795, |
| "abs_max": 5.21875, |
| "frac_dead": 6.329571624519303e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.proj": { |
| "mean": 0.007564673200249672, |
| "std": 0.4087885618209839, |
| "abs_max": 3.0, |
| "frac_dead": 0.00010850694525288418, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.0.op": { |
| "mean": -0.17015184462070465, |
| "std": 1.6812559366226196, |
| "abs_max": 11.4375, |
| "frac_dead": 9.960598617908545e-06, |
| "frac_sat": 3.496805948088877e-05 |
| }, |
| "dec_blocks.1.norm1": { |
| "mean": -0.00024216788006015122, |
| "std": 1.0003820657730103, |
| "abs_max": 9.206901550292969, |
| "frac_dead": 3.178914482759865e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.conv1": { |
| "mean": -1.724596381187439, |
| "std": 3.3776965141296387, |
| "abs_max": 57.5, |
| "frac_dead": 4.662407718569739e-06, |
| "frac_sat": 0.03395335003733635 |
| }, |
| "dec_blocks.1.t_proj": { |
| "mean": -0.0024088711943477392, |
| "std": 0.07598824799060822, |
| "abs_max": 0.49609375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.norm2": { |
| "mean": -0.00754527747631073, |
| "std": 0.9961175918579102, |
| "abs_max": 14.618008613586426, |
| "frac_dead": 4.238552719471045e-06, |
| "frac_sat": 5.997551852487959e-05 |
| }, |
| "dec_blocks.1.conv2": { |
| "mean": -9.273696377931628e-06, |
| "std": 0.6917805075645447, |
| "abs_max": 4.25, |
| "frac_dead": 2.47955322265625e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.skip": { |
| "mean": -0.006895574275404215, |
| "std": 0.9343755841255188, |
| "abs_max": 6.59375, |
| "frac_dead": 3.687540811370127e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.norm": { |
| "mean": 1.738758510327898e-05, |
| "std": 0.9983002543449402, |
| "abs_max": 6.872876167297363, |
| "frac_dead": 2.119276274470394e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.qkv": { |
| "mean": -0.0012654407182708383, |
| "std": 0.6738247871398926, |
| "abs_max": 5.1875, |
| "frac_dead": 6.824069714639336e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.proj": { |
| "mean": -0.007716475520282984, |
| "std": 0.3779028654098511, |
| "abs_max": 2.96875, |
| "frac_dead": 0.00011444091796875, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.1.op": { |
| "mean": -0.013204764574766159, |
| "std": 1.2440454959869385, |
| "abs_max": 8.4375, |
| "frac_dead": 1.2980567589693237e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm1": { |
| "mean": -0.00021323782857507467, |
| "std": 1.000333309173584, |
| "abs_max": 8.981050491333008, |
| "frac_dead": 8.830317597130488e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.conv1": { |
| "mean": -0.8999322056770325, |
| "std": 2.557943105697632, |
| "abs_max": 42.0, |
| "frac_dead": 1.3457404747896362e-05, |
| "frac_sat": 0.018307263031601906 |
| }, |
| "dec_blocks.2.t_proj": { |
| "mean": -0.00454047741368413, |
| "std": 0.08159489184617996, |
| "abs_max": 0.3828125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm2": { |
| "mean": -0.008707795292139053, |
| "std": 0.9994327425956726, |
| "abs_max": 13.457539558410645, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 3.030565130757168e-05 |
| }, |
| "dec_blocks.2.conv2": { |
| "mean": 0.012484816834330559, |
| "std": 0.3138270080089569, |
| "abs_max": 1.7265625, |
| "frac_dead": 6.633334851358086e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.skip": { |
| "mean": 0.02413799613714218, |
| "std": 1.0058594942092896, |
| "abs_max": 6.75, |
| "frac_dead": 3.751118856598623e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.norm": { |
| "mean": -6.64558683638461e-05, |
| "std": 0.997047483921051, |
| "abs_max": 6.378185272216797, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.qkv": { |
| "mean": 0.004041221924126148, |
| "std": 0.6196464896202087, |
| "abs_max": 5.3125, |
| "frac_dead": 9.222383960150182e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.proj": { |
| "mean": -0.003088516416028142, |
| "std": 0.3288596570491791, |
| "abs_max": 2.625, |
| "frac_dead": 0.00018077426648233086, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.2.op": { |
| "mean": 0.06600978970527649, |
| "std": 0.8171164393424988, |
| "abs_max": 5.96875, |
| "frac_dead": 2.7126736313221045e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm1": { |
| "mean": -3.3551619708305225e-05, |
| "std": 1.0003398656845093, |
| "abs_max": 7.884114742279053, |
| "frac_dead": 9.271833505408722e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.conv1": { |
| "mean": -0.808504045009613, |
| "std": 1.9701484441757202, |
| "abs_max": 30.25, |
| "frac_dead": 1.796086689864751e-05, |
| "frac_sat": 0.003796603996306658 |
| }, |
| "dec_blocks.3.t_proj": { |
| "mean": -0.000398904929170385, |
| "std": 0.08900758624076843, |
| "abs_max": 0.48046875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm2": { |
| "mean": -0.008217168040573597, |
| "std": 0.9956994652748108, |
| "abs_max": 11.516836166381836, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 4.212061412545154e-06 |
| }, |
| "dec_blocks.3.conv2": { |
| "mean": 0.010301654227077961, |
| "std": 0.3566579222679138, |
| "abs_max": 3.125, |
| "frac_dead": 5.910131585551426e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.skip": { |
| "mean": -0.017353812232613564, |
| "std": 0.6160323619842529, |
| "abs_max": 4.15625, |
| "frac_dead": 7.367134094238281e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.3.op": { |
| "mean": -0.02672424353659153, |
| "std": 0.6458936929702759, |
| "abs_max": 7.09375, |
| "frac_dead": 3.521309918141924e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm1": { |
| "mean": -0.00045055372174829245, |
| "std": 0.9998191595077515, |
| "abs_max": 10.019762992858887, |
| "frac_dead": 7.108405952749308e-07, |
| "frac_sat": 4.4151589051466544e-09 |
| }, |
| "dec_blocks.4.conv1": { |
| "mean": -0.7278802990913391, |
| "std": 1.838720440864563, |
| "abs_max": 19.625, |
| "frac_dead": 1.2795130714948755e-05, |
| "frac_sat": 0.000757283647544682 |
| }, |
| "dec_blocks.4.t_proj": { |
| "mean": -0.003106130985543132, |
| "std": 0.09513773769140244, |
| "abs_max": 0.453125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm2": { |
| "mean": -0.0046213483437895775, |
| "std": 0.996771514415741, |
| "abs_max": 10.407454490661621, |
| "frac_dead": 4.635916752704361e-07, |
| "frac_sat": 1.059638137235197e-07 |
| }, |
| "dec_blocks.4.conv2": { |
| "mean": -0.05116056278347969, |
| "std": 0.52388596534729, |
| "abs_max": 6.03125, |
| "frac_dead": 7.355213165283203e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.skip": { |
| "mean": -0.051753975450992584, |
| "std": 0.7470393776893616, |
| "abs_max": 6.25, |
| "frac_dead": 7.97775064711459e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.4.op": { |
| "mean": -0.22740226984024048, |
| "std": 1.3430253267288208, |
| "abs_max": 18.5, |
| "frac_dead": 3.66667918569874e-05, |
| "frac_sat": 0.00016431344556622207 |
| }, |
| "dec_blocks.5.norm1": { |
| "mean": -0.0034408338833600283, |
| "std": 0.9988682866096497, |
| "abs_max": 13.417811393737793, |
| "frac_dead": 1.0381141919424408e-06, |
| "frac_sat": 1.564621925354004e-06 |
| }, |
| "dec_blocks.5.conv1": { |
| "mean": -0.1277112513780594, |
| "std": 1.29868483543396, |
| "abs_max": 10.4375, |
| "frac_dead": 1.688467091298662e-05, |
| "frac_sat": 1.655684656043377e-08 |
| }, |
| "dec_blocks.5.t_proj": { |
| "mean": 0.009061946533620358, |
| "std": 0.16928990185260773, |
| "abs_max": 1.0859375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.norm2": { |
| "mean": -0.00034307813621126115, |
| "std": 1.0034955739974976, |
| "abs_max": 9.536330223083496, |
| "frac_dead": 5.46375929388887e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.conv2": { |
| "mean": -0.30506497621536255, |
| "std": 1.493102788925171, |
| "abs_max": 17.875, |
| "frac_dead": 2.8067164748790674e-05, |
| "frac_sat": 0.00032416978501714766 |
| }, |
| "dec_blocks.5.skip": { |
| "mean": -0.15524762868881226, |
| "std": 1.0975347757339478, |
| "abs_max": 10.1875, |
| "frac_dead": 6.643599772360176e-05, |
| "frac_sat": 2.317958447406454e-08 |
| }, |
| "out_norm": { |
| "mean": -0.00539039820432663, |
| "std": 0.9967310428619385, |
| "abs_max": 10.310091972351074, |
| "frac_dead": 3.675619666410057e-07, |
| "frac_sat": 1.3245476715439963e-08 |
| }, |
| "out_conv": { |
| "mean": 0.1330542266368866, |
| "std": 0.20147153735160828, |
| "abs_max": 0.96484375, |
| "frac_dead": 2.3312039047596045e-05, |
| "frac_sat": 0.0 |
| } |
| } |
| } |
