| { |
| "iter": 1000, |
| "per_layer": { |
| "time_mlp.1.weight": { |
| "w_norm": 18.69807243347168, |
| "g_norm": 0.0460389219224453, |
| "shape": [ |
| 1024, |
| 256 |
| ] |
| }, |
| "time_mlp.1.bias": { |
| "w_norm": 1.1588034629821777, |
| "g_norm": 0.004071285482496023, |
| "shape": [ |
| 1024 |
| ] |
| }, |
| "time_mlp.3.weight": { |
| "w_norm": 9.50298023223877, |
| "g_norm": 0.018244950100779533, |
| "shape": [ |
| 256, |
| 1024 |
| ] |
| }, |
| "time_mlp.3.bias": { |
| "w_norm": 0.2909029722213745, |
| "g_norm": 0.005334504880011082, |
| "shape": [ |
| 256 |
| ] |
| }, |
| "input_conv.weight": { |
| "w_norm": 7.937385082244873, |
| "g_norm": 0.04928673058748245, |
| "shape": [ |
| 192, |
| 6, |
| 3, |
| 3 |
| ] |
| }, |
| "input_conv.bias": { |
| "w_norm": 1.1879003047943115, |
| "g_norm": 0.06297601759433746, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.weight": { |
| "w_norm": 13.822595596313477, |
| "g_norm": 0.016159730032086372, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.bias": { |
| "w_norm": 0.06115815043449402, |
| "g_norm": 0.024567998945713043, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv1.weight": { |
| "w_norm": 8.222108840942383, |
| "g_norm": 0.1748814433813095, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv1.bias": { |
| "w_norm": 0.20211568474769592, |
| "g_norm": 0.01814374327659607, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.t_proj.weight": { |
| "w_norm": 8.027454376220703, |
| "g_norm": 0.008248274214565754, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.0.t_proj.bias": { |
| "w_norm": 0.5023583173751831, |
| "g_norm": 0.018131105229258537, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.weight": { |
| "w_norm": 13.845124244689941, |
| "g_norm": 0.006467544473707676, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.bias": { |
| "w_norm": 0.035126347094774246, |
| "g_norm": 0.00743205426260829, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv2.weight": { |
| "w_norm": 8.169148445129395, |
| "g_norm": 0.07441551238298416, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv2.bias": { |
| "w_norm": 0.1873043030500412, |
| "g_norm": 0.005247351247817278, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.weight": { |
| "w_norm": 13.851595878601074, |
| "g_norm": 3.0715102184331045e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.bias": { |
| "w_norm": 0.019507592543959618, |
| "g_norm": 2.6044943297165446e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv1.weight": { |
| "w_norm": 8.134106636047363, |
| "g_norm": 0.0014839934883639216, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv1.bias": { |
| "w_norm": 0.19206438958644867, |
| "g_norm": 6.173425208544359e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.t_proj.weight": { |
| "w_norm": 8.007999420166016, |
| "g_norm": 3.269246008130722e-05, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.1.t_proj.bias": { |
| "w_norm": 0.5217474102973938, |
| "g_norm": 6.174553709570318e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.weight": { |
| "w_norm": 13.828197479248047, |
| "g_norm": 0.00014807471598032862, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.bias": { |
| "w_norm": 0.031187444925308228, |
| "g_norm": 0.00011828799324575812, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv2.weight": { |
| "w_norm": 8.091903686523438, |
| "g_norm": 0.004907071590423584, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv2.bias": { |
| "w_norm": 0.20746883749961853, |
| "g_norm": 0.0002124477905454114, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.weight": { |
| "w_norm": 13.850759506225586, |
| "g_norm": 9.877112461254e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.bias": { |
| "w_norm": 0.027128858491778374, |
| "g_norm": 9.442120062885806e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.conv1.weight": { |
| "w_norm": 11.504433631896973, |
| "g_norm": 0.00176367093808949, |
| "shape": [ |
| 384, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv1.bias": { |
| "w_norm": 0.2661369740962982, |
| "g_norm": 7.727985212113708e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.t_proj.weight": { |
| "w_norm": 11.387003898620605, |
| "g_norm": 4.298925705370493e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.2.t_proj.bias": { |
| "w_norm": 0.7206763625144958, |
| "g_norm": 7.726851617917418e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.weight": { |
| "w_norm": 19.513442993164062, |
| "g_norm": 6.003796079312451e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.bias": { |
| "w_norm": 0.07231348007917404, |
| "g_norm": 5.830789086758159e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.conv2.weight": { |
| "w_norm": 11.476949691772461, |
| "g_norm": 0.002501599956303835, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv2.bias": { |
| "w_norm": 0.2027990072965622, |
| "g_norm": 0.00010226813901681453, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.skip.weight": { |
| "w_norm": 11.265475273132324, |
| "g_norm": 0.0029775125440210104, |
| "shape": [ |
| 384, |
| 192, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.2.skip.bias": { |
| "w_norm": 0.8253466486930847, |
| "g_norm": 0.00010226813901681453, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.weight": { |
| "w_norm": 19.597091674804688, |
| "g_norm": 7.468698640877847e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.bias": { |
| "w_norm": 0.03287740424275398, |
| "g_norm": 5.63786943530431e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv1.weight": { |
| "w_norm": 11.572396278381348, |
| "g_norm": 0.00018376416119281203, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv1.bias": { |
| "w_norm": 0.19801898300647736, |
| "g_norm": 5.174441412236774e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.t_proj.weight": { |
| "w_norm": 11.330994606018066, |
| "g_norm": 2.8422682589734904e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.3.t_proj.bias": { |
| "w_norm": 0.690720796585083, |
| "g_norm": 5.173218596610241e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.weight": { |
| "w_norm": 19.522457122802734, |
| "g_norm": 1.5316471035475843e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.bias": { |
| "w_norm": 0.07057540118694305, |
| "g_norm": 7.84625080996193e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv2.weight": { |
| "w_norm": 11.416537284851074, |
| "g_norm": 0.0004853545979131013, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv2.bias": { |
| "w_norm": 0.19048988819122314, |
| "g_norm": 1.8962364265462384e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.weight": { |
| "w_norm": 19.59347915649414, |
| "g_norm": 3.5858311093761586e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.bias": { |
| "w_norm": 0.03173157572746277, |
| "g_norm": 4.146143055550056e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.conv1.weight": { |
| "w_norm": 16.279399871826172, |
| "g_norm": 0.0001275187823921442, |
| "shape": [ |
| 768, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv1.bias": { |
| "w_norm": 0.28021085262298584, |
| "g_norm": 4.764061486639548e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.t_proj.weight": { |
| "w_norm": 16.028104782104492, |
| "g_norm": 2.406797420917428e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.4.t_proj.bias": { |
| "w_norm": 0.990105926990509, |
| "g_norm": 4.763509878102923e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.weight": { |
| "w_norm": 27.602861404418945, |
| "g_norm": 6.712813046760857e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.bias": { |
| "w_norm": 0.10398496687412262, |
| "g_norm": 7.564138741145143e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.conv2.weight": { |
| "w_norm": 16.15690803527832, |
| "g_norm": 0.00038142167613841593, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv2.bias": { |
| "w_norm": 0.19474577903747559, |
| "g_norm": 1.245525709236972e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.skip.weight": { |
| "w_norm": 15.938165664672852, |
| "g_norm": 0.0006917706923559308, |
| "shape": [ |
| 768, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.4.skip.bias": { |
| "w_norm": 0.8077543377876282, |
| "g_norm": 1.245525709236972e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.weight": { |
| "w_norm": 27.713071823120117, |
| "g_norm": 8.311205306199554e-07, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.bias": { |
| "w_norm": 0.05238156393170357, |
| "g_norm": 1.1990509847237263e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv1.weight": { |
| "w_norm": 16.445384979248047, |
| "g_norm": 5.676236833096482e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv1.bias": { |
| "w_norm": 0.20182201266288757, |
| "g_norm": 1.51737867781776e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.t_proj.weight": { |
| "w_norm": 15.997159957885742, |
| "g_norm": 7.934851851132407e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.5.t_proj.bias": { |
| "w_norm": 0.9971441030502319, |
| "g_norm": 1.5173413885349873e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.weight": { |
| "w_norm": 27.63616943359375, |
| "g_norm": 2.516604354241281e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.bias": { |
| "w_norm": 0.08157779276371002, |
| "g_norm": 3.0129847345961025e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv2.weight": { |
| "w_norm": 16.266035079956055, |
| "g_norm": 0.00022640718088950962, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv2.bias": { |
| "w_norm": 0.19263482093811035, |
| "g_norm": 7.249552709254203e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.3.norm.weight": { |
| "w_norm": 19.5748291015625, |
| "g_norm": 7.831244147382677e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.norm.bias": { |
| "w_norm": 0.041913531720638275, |
| "g_norm": 7.98945166025078e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.qkv.weight": { |
| "w_norm": 19.660964965820312, |
| "g_norm": 0.00023820465139579028, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.qkv.bias": { |
| "w_norm": 1.0083452463150024, |
| "g_norm": 1.2227897968841717e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "enc_attns.3.proj.weight": { |
| "w_norm": 11.306562423706055, |
| "g_norm": 0.00021490734070539474, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.proj.bias": { |
| "w_norm": 0.5660514235496521, |
| "g_norm": 1.9022467313334346e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.4.norm.weight": { |
| "w_norm": 27.66777229309082, |
| "g_norm": 2.031427129622898e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.norm.bias": { |
| "w_norm": 0.034273792058229446, |
| "g_norm": 4.666352651838679e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.qkv.weight": { |
| "w_norm": 27.864606857299805, |
| "g_norm": 8.733879076316953e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.qkv.bias": { |
| "w_norm": 1.0126442909240723, |
| "g_norm": 7.294429906323785e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.4.proj.weight": { |
| "w_norm": 15.98458194732666, |
| "g_norm": 7.8769022366032e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.proj.bias": { |
| "w_norm": 0.5737269520759583, |
| "g_norm": 1.2380953194224276e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.weight": { |
| "w_norm": 27.681488037109375, |
| "g_norm": 1.4364586604642682e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.bias": { |
| "w_norm": 0.03207901865243912, |
| "g_norm": 2.4433634280285332e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.qkv.weight": { |
| "w_norm": 27.812726974487305, |
| "g_norm": 6.495189882116392e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.qkv.bias": { |
| "w_norm": 0.9992081522941589, |
| "g_norm": 4.105900643480709e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.5.proj.weight": { |
| "w_norm": 16.01068687438965, |
| "g_norm": 7.401613402180374e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.proj.bias": { |
| "w_norm": 0.5965924859046936, |
| "g_norm": 7.23396624380257e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_downs.0.op.weight": { |
| "w_norm": 8.119194030761719, |
| "g_norm": 0.00474424147978425, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.0.op.bias": { |
| "w_norm": 0.18655341863632202, |
| "g_norm": 0.0002124822058249265, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.1.op.weight": { |
| "w_norm": 8.064340591430664, |
| "g_norm": 0.00629104720428586, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.1.op.bias": { |
| "w_norm": 0.19401651620864868, |
| "g_norm": 0.00010197861411143094, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.2.op.weight": { |
| "w_norm": 11.412195205688477, |
| "g_norm": 0.0016165502602234483, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.2.op.bias": { |
| "w_norm": 0.18826229870319366, |
| "g_norm": 1.9014922145288438e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.3.op.weight": { |
| "w_norm": 11.37773609161377, |
| "g_norm": 0.0009675542241893709, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.3.op.bias": { |
| "w_norm": 0.19311626255512238, |
| "g_norm": 8.070419426076114e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.4.op.weight": { |
| "w_norm": 16.181838989257812, |
| "g_norm": 0.0011415004264563322, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.4.op.bias": { |
| "w_norm": 0.19675636291503906, |
| "g_norm": 7.221314263006207e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.weight": { |
| "w_norm": 27.711105346679688, |
| "g_norm": 1.734056468194467e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.bias": { |
| "w_norm": 0.04435775801539421, |
| "g_norm": 1.6670099967086571e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv1.weight": { |
| "w_norm": 16.44451332092285, |
| "g_norm": 6.236001354409382e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv1.bias": { |
| "w_norm": 0.20257723331451416, |
| "g_norm": 1.5994581872291747e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.t_proj.weight": { |
| "w_norm": 16.007741928100586, |
| "g_norm": 8.396062298743345e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid1.t_proj.bias": { |
| "w_norm": 1.0318254232406616, |
| "g_norm": 1.5996096180970198e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.weight": { |
| "w_norm": 27.638864517211914, |
| "g_norm": 2.9384184472291963e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.bias": { |
| "w_norm": 0.08424504101276398, |
| "g_norm": 3.2435036700917408e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv2.weight": { |
| "w_norm": 16.264646530151367, |
| "g_norm": 0.00015367653395514935, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv2.bias": { |
| "w_norm": 0.19507506489753723, |
| "g_norm": 5.045553280069726e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.weight": { |
| "w_norm": 27.687448501586914, |
| "g_norm": 1.162366174867202e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.bias": { |
| "w_norm": 0.029944565147161484, |
| "g_norm": 1.7182350120492629e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.qkv.weight": { |
| "w_norm": 27.804889678955078, |
| "g_norm": 5.264316496322863e-05, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.qkv.bias": { |
| "w_norm": 0.9861231446266174, |
| "g_norm": 2.871294555006898e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "mid_attn.proj.weight": { |
| "w_norm": 16.009910583496094, |
| "g_norm": 5.811886876472272e-05, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.proj.bias": { |
| "w_norm": 0.576531708240509, |
| "g_norm": 5.044004410592606e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.weight": { |
| "w_norm": 27.714092254638672, |
| "g_norm": 2.8204319733049488e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.bias": { |
| "w_norm": 0.04821281507611275, |
| "g_norm": 2.875601012419793e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv1.weight": { |
| "w_norm": 16.43076515197754, |
| "g_norm": 7.036297029117122e-05, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv1.bias": { |
| "w_norm": 0.19785825908184052, |
| "g_norm": 1.8724236952039064e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.t_proj.weight": { |
| "w_norm": 15.99503231048584, |
| "g_norm": 9.8669033832266e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid2.t_proj.bias": { |
| "w_norm": 0.9966008067131042, |
| "g_norm": 1.8730396504906821e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.weight": { |
| "w_norm": 27.629024505615234, |
| "g_norm": 4.045460173074389e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.bias": { |
| "w_norm": 0.09099399298429489, |
| "g_norm": 3.6871888369205408e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv2.weight": { |
| "w_norm": 16.222124099731445, |
| "g_norm": 0.00016647143638692796, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv2.bias": { |
| "w_norm": 0.189836323261261, |
| "g_norm": 5.019630407332443e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm1.weight": { |
| "w_norm": 39.19388198852539, |
| "g_norm": 1.2530965705082053e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.norm1.bias": { |
| "w_norm": 0.05807539448142052, |
| "g_norm": 1.1799430694736657e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.conv1.weight": { |
| "w_norm": 16.877586364746094, |
| "g_norm": 7.809244561940432e-05, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv1.bias": { |
| "w_norm": 0.14163629710674286, |
| "g_norm": 1.5247343299051863e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.t_proj.weight": { |
| "w_norm": 16.0123348236084, |
| "g_norm": 7.94084598965128e-07, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.0.t_proj.bias": { |
| "w_norm": 1.0203391313552856, |
| "g_norm": 1.525260472590162e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.weight": { |
| "w_norm": 27.59062385559082, |
| "g_norm": 2.9507796170946676e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.bias": { |
| "w_norm": 0.11849326640367508, |
| "g_norm": 4.340486611909e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.conv2.weight": { |
| "w_norm": 16.1192626953125, |
| "g_norm": 0.00026150690973736346, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv2.bias": { |
| "w_norm": 0.19648997485637665, |
| "g_norm": 1.0889924851653632e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.skip.weight": { |
| "w_norm": 15.902674674987793, |
| "g_norm": 0.001218110672198236, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.0.skip.bias": { |
| "w_norm": 0.41060250997543335, |
| "g_norm": 1.0889924851653632e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm1.weight": { |
| "w_norm": 39.1891975402832, |
| "g_norm": 9.350132131658029e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.norm1.bias": { |
| "w_norm": 0.053357407450675964, |
| "g_norm": 1.0519148418097757e-05, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.conv1.weight": { |
| "w_norm": 16.819643020629883, |
| "g_norm": 0.00029541790718212724, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv1.bias": { |
| "w_norm": 0.13914765417575836, |
| "g_norm": 5.622245680569904e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.t_proj.weight": { |
| "w_norm": 16.038969039916992, |
| "g_norm": 2.8538920560094994e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.1.t_proj.bias": { |
| "w_norm": 0.9858460426330566, |
| "g_norm": 5.6221151680801995e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.weight": { |
| "w_norm": 27.590665817260742, |
| "g_norm": 9.302767466579098e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.bias": { |
| "w_norm": 0.12908512353897095, |
| "g_norm": 1.4394856407307088e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.conv2.weight": { |
| "w_norm": 16.1109619140625, |
| "g_norm": 0.0006568268872797489, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv2.bias": { |
| "w_norm": 0.19586755335330963, |
| "g_norm": 2.2840289602754638e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.skip.weight": { |
| "w_norm": 15.894036293029785, |
| "g_norm": 0.0016062397044152021, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.1.skip.bias": { |
| "w_norm": 0.40641382336616516, |
| "g_norm": 2.2840289602754638e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.2.norm1.weight": { |
| "w_norm": 33.943851470947266, |
| "g_norm": 1.3375498383538797e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.norm1.bias": { |
| "w_norm": 0.05633563548326492, |
| "g_norm": 1.2195808267279062e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.conv1.weight": { |
| "w_norm": 11.906632423400879, |
| "g_norm": 0.0004902135115116835, |
| "shape": [ |
| 384, |
| 1152, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv1.bias": { |
| "w_norm": 0.11643993854522705, |
| "g_norm": 1.0101834959641565e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.t_proj.weight": { |
| "w_norm": 11.322545051574707, |
| "g_norm": 5.009514097764622e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.2.t_proj.bias": { |
| "w_norm": 0.7094705700874329, |
| "g_norm": 1.0103261047333945e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.weight": { |
| "w_norm": 19.51650619506836, |
| "g_norm": 2.3810951461200602e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.bias": { |
| "w_norm": 0.085000179708004, |
| "g_norm": 3.0295035685412586e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.conv2.weight": { |
| "w_norm": 11.389786720275879, |
| "g_norm": 0.0011060911929234862, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv2.bias": { |
| "w_norm": 0.18991783261299133, |
| "g_norm": 5.410888115875423e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.skip.weight": { |
| "w_norm": 11.173041343688965, |
| "g_norm": 0.0025957406032830477, |
| "shape": [ |
| 384, |
| 1152, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.2.skip.bias": { |
| "w_norm": 0.3321686089038849, |
| "g_norm": 5.410888115875423e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm1.weight": { |
| "w_norm": 27.71334457397461, |
| "g_norm": 6.114856660133228e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.norm1.bias": { |
| "w_norm": 0.05490655079483986, |
| "g_norm": 5.5085638450691476e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.conv1.weight": { |
| "w_norm": 11.796333312988281, |
| "g_norm": 0.0024704246316105127, |
| "shape": [ |
| 384, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv1.bias": { |
| "w_norm": 0.14261765778064728, |
| "g_norm": 5.7586727052694187e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.t_proj.weight": { |
| "w_norm": 11.342194557189941, |
| "g_norm": 2.818637221935205e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.3.t_proj.bias": { |
| "w_norm": 0.7088264226913452, |
| "g_norm": 5.7579698477638885e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.weight": { |
| "w_norm": 19.502674102783203, |
| "g_norm": 6.80582452332601e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.bias": { |
| "w_norm": 0.08563433587551117, |
| "g_norm": 8.817730122245848e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.conv2.weight": { |
| "w_norm": 11.395421981811523, |
| "g_norm": 0.004227672703564167, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv2.bias": { |
| "w_norm": 0.19376030564308167, |
| "g_norm": 0.00019759820133913308, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.skip.weight": { |
| "w_norm": 11.19637680053711, |
| "g_norm": 0.006898099556565285, |
| "shape": [ |
| 384, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.3.skip.bias": { |
| "w_norm": 0.40675294399261475, |
| "g_norm": 0.00019759820133913308, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.4.norm1.weight": { |
| "w_norm": 23.995790481567383, |
| "g_norm": 7.700704009039328e-05, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.norm1.bias": { |
| "w_norm": 0.05365603044629097, |
| "g_norm": 7.204065332189202e-05, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.conv1.weight": { |
| "w_norm": 8.276124000549316, |
| "g_norm": 0.0042354753240942955, |
| "shape": [ |
| 192, |
| 576, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv1.bias": { |
| "w_norm": 0.11667870730161667, |
| "g_norm": 0.0001092359161702916, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.t_proj.weight": { |
| "w_norm": 8.005839347839355, |
| "g_norm": 5.376513217925094e-05, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.4.t_proj.bias": { |
| "w_norm": 0.4811122417449951, |
| "g_norm": 0.00010927868424914777, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.weight": { |
| "w_norm": 13.81904125213623, |
| "g_norm": 0.000297811086056754, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.bias": { |
| "w_norm": 0.03246624395251274, |
| "g_norm": 0.0002632128889672458, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.conv2.weight": { |
| "w_norm": 8.034357070922852, |
| "g_norm": 0.010364442132413387, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv2.bias": { |
| "w_norm": 0.19388015568256378, |
| "g_norm": 0.0004662639112211764, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.skip.weight": { |
| "w_norm": 7.931014060974121, |
| "g_norm": 0.013633654452860355, |
| "shape": [ |
| 192, |
| 576, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.4.skip.bias": { |
| "w_norm": 0.3468608558177948, |
| "g_norm": 0.0004662639112211764, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm1.weight": { |
| "w_norm": 19.574352264404297, |
| "g_norm": 0.002302073873579502, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.norm1.bias": { |
| "w_norm": 0.05974191799759865, |
| "g_norm": 0.002887281822040677, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.conv1.weight": { |
| "w_norm": 8.306647300720215, |
| "g_norm": 0.05203191936016083, |
| "shape": [ |
| 192, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv1.bias": { |
| "w_norm": 0.13815189898014069, |
| "g_norm": 0.00202044565230608, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.t_proj.weight": { |
| "w_norm": 8.092968940734863, |
| "g_norm": 0.000958150252699852, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.5.t_proj.bias": { |
| "w_norm": 0.5047914981842041, |
| "g_norm": 0.002020731568336487, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.weight": { |
| "w_norm": 13.871760368347168, |
| "g_norm": 0.0022596679627895355, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.bias": { |
| "w_norm": 0.030323660001158714, |
| "g_norm": 0.0025028937961906195, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.conv2.weight": { |
| "w_norm": 8.181824684143066, |
| "g_norm": 0.060832276940345764, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv2.bias": { |
| "w_norm": 0.1982647031545639, |
| "g_norm": 0.00286358711309731, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.skip.weight": { |
| "w_norm": 8.064077377319336, |
| "g_norm": 0.07279777526855469, |
| "shape": [ |
| 192, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.5.skip.bias": { |
| "w_norm": 0.42116665840148926, |
| "g_norm": 0.00286358711309731, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_attns.0.norm.weight": { |
| "w_norm": 27.669570922851562, |
| "g_norm": 3.4397162380628288e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.norm.bias": { |
| "w_norm": 0.024704044684767723, |
| "g_norm": 4.100290880160173e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.qkv.weight": { |
| "w_norm": 27.809728622436523, |
| "g_norm": 0.00014507168089039624, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.qkv.bias": { |
| "w_norm": 1.0090502500534058, |
| "g_norm": 6.295892490015831e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.0.proj.weight": { |
| "w_norm": 16.01603126525879, |
| "g_norm": 0.0001487144036218524, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.proj.bias": { |
| "w_norm": 0.5765091776847839, |
| "g_norm": 1.0640424989105668e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.weight": { |
| "w_norm": 27.69176483154297, |
| "g_norm": 9.369830877403729e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.bias": { |
| "w_norm": 0.03322061523795128, |
| "g_norm": 1.1450497368059587e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.qkv.weight": { |
| "w_norm": 27.846345901489258, |
| "g_norm": 0.0003313145716674626, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.qkv.bias": { |
| "w_norm": 1.0051014423370361, |
| "g_norm": 1.5137110494833905e-05, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.1.proj.weight": { |
| "w_norm": 16.0462703704834, |
| "g_norm": 0.00032928746077232063, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.proj.bias": { |
| "w_norm": 0.5801531076431274, |
| "g_norm": 2.3453705580323003e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.2.norm.weight": { |
| "w_norm": 19.559301376342773, |
| "g_norm": 1.526609958091285e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.norm.bias": { |
| "w_norm": 0.01634746603667736, |
| "g_norm": 1.7435884728911333e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.qkv.weight": { |
| "w_norm": 19.649356842041016, |
| "g_norm": 0.0005132962833158672, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.qkv.bias": { |
| "w_norm": 0.987482488155365, |
| "g_norm": 2.9351922421483323e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_attns.2.proj.weight": { |
| "w_norm": 11.306385040283203, |
| "g_norm": 0.0005538715049624443, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.proj.bias": { |
| "w_norm": 0.5871843695640564, |
| "g_norm": 5.271810368867591e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.0.op.weight": { |
| "w_norm": 16.088212966918945, |
| "g_norm": 0.0013465607771649957, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.0.op.bias": { |
| "w_norm": 0.2013898640871048, |
| "g_norm": 1.0867960554605816e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.1.op.weight": { |
| "w_norm": 16.08967399597168, |
| "g_norm": 0.002522860188037157, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.1.op.bias": { |
| "w_norm": 0.1886485069990158, |
| "g_norm": 2.7130645321449265e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.2.op.weight": { |
| "w_norm": 11.30615520477295, |
| "g_norm": 0.006385178305208683, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.2.op.bias": { |
| "w_norm": 0.19624017179012299, |
| "g_norm": 9.016467083711177e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.3.op.weight": { |
| "w_norm": 11.316301345825195, |
| "g_norm": 0.014089048840105534, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.3.op.bias": { |
| "w_norm": 0.19712364673614502, |
| "g_norm": 0.00023156803217716515, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.4.op.weight": { |
| "w_norm": 7.999128818511963, |
| "g_norm": 0.03885931521654129, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.4.op.bias": { |
| "w_norm": 0.19474607706069946, |
| "g_norm": 0.0006655824836343527, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.weight": { |
| "w_norm": 13.822661399841309, |
| "g_norm": 0.005152558907866478, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.bias": { |
| "w_norm": 0.04736120253801346, |
| "g_norm": 0.008335559628903866, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_conv.weight": { |
| "w_norm": 0.11635566502809525, |
| "g_norm": 0.9671086668968201, |
| "shape": [ |
| 3, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "out_conv.bias": { |
| "w_norm": 0.007038481067866087, |
| "g_norm": 0.06203581020236015, |
| "shape": [ |
| 3 |
| ] |
| } |
| }, |
| "activations": { |
| "input_conv": { |
| "mean": 0.0005878935335204005, |
| "std": 0.3478555381298065, |
| "abs_max": 2.53125, |
| "frac_dead": 0.0005414850311353803, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.1": { |
| "mean": 0.02260744199156761, |
| "std": 0.27119144797325134, |
| "abs_max": 3.0625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.3": { |
| "mean": 0.019019395112991333, |
| "std": 0.13746215403079987, |
| "abs_max": 0.59765625, |
| "frac_dead": 0.00048828125, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm1": { |
| "mean": -0.0034768523182719946, |
| "std": 0.9972635507583618, |
| "abs_max": 6.692741870880127, |
| "frac_dead": 9.437401899958786e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv1": { |
| "mean": -0.03561967611312866, |
| "std": 0.5352197885513306, |
| "abs_max": 7.875, |
| "frac_dead": 6.592604768229648e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.t_proj": { |
| "mean": -0.0022988717537373304, |
| "std": 0.10281845182180405, |
| "abs_max": 0.71484375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm2": { |
| "mean": -0.0019118194468319416, |
| "std": 0.9995514750480652, |
| "abs_max": 10.405152320861816, |
| "frac_dead": 1.1788474694185425e-06, |
| "frac_sat": 6.622738357719982e-09 |
| }, |
| "enc_blocks.0.conv2": { |
| "mean": -0.009182874113321304, |
| "std": 0.8037123680114746, |
| "abs_max": 7.125, |
| "frac_dead": 3.254082548664883e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.0.op": { |
| "mean": -0.1348954737186432, |
| "std": 1.5058555603027344, |
| "abs_max": 10.5625, |
| "frac_dead": 1.3192494407121558e-05, |
| "frac_sat": 1.3245476715439963e-08 |
| }, |
| "enc_blocks.1.norm1": { |
| "mean": -0.0008627756033092737, |
| "std": 0.9998546838760376, |
| "abs_max": 6.996696472167969, |
| "frac_dead": 2.914004824106087e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv1": { |
| "mean": -0.21788886189460754, |
| "std": 1.1535621881484985, |
| "abs_max": 7.21875, |
| "frac_dead": 2.4888251573429443e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.t_proj": { |
| "mean": 0.0034813284873962402, |
| "std": 0.06453254073858261, |
| "abs_max": 0.322265625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.norm2": { |
| "mean": -0.0024751583114266396, |
| "std": 0.9978860020637512, |
| "abs_max": 7.0123209953308105, |
| "frac_dead": 5.563099989558395e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv2": { |
| "mean": 0.009306020103394985, |
| "std": 0.946929931640625, |
| "abs_max": 7.53125, |
| "frac_dead": 4.3299463868606836e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.1.op": { |
| "mean": -0.30516207218170166, |
| "std": 2.371323585510254, |
| "abs_max": 13.0625, |
| "frac_dead": 1.1973910659435205e-05, |
| "frac_sat": 6.0664282500511035e-05 |
| }, |
| "enc_blocks.2.norm1": { |
| "mean": -0.0008903791313059628, |
| "std": 0.9997683763504028, |
| "abs_max": 6.303738117218018, |
| "frac_dead": 1.271565793103946e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.conv1": { |
| "mean": -0.265188604593277, |
| "std": 0.748349130153656, |
| "abs_max": 6.1875, |
| "frac_dead": 4.259745401213877e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.t_proj": { |
| "mean": -0.0014002572279423475, |
| "std": 0.07900581508874893, |
| "abs_max": 0.53125, |
| "frac_dead": 0.00010850694525288418, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.norm2": { |
| "mean": -0.0040403506718575954, |
| "std": 0.9962407946586609, |
| "abs_max": 6.904444694519043, |
| "frac_dead": 6.092919306865952e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.conv2": { |
| "mean": 0.13902650773525238, |
| "std": 0.9115206599235535, |
| "abs_max": 9.6875, |
| "frac_dead": 2.763006523309741e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.skip": { |
| "mean": 0.06025930121541023, |
| "std": 1.4612077474594116, |
| "abs_max": 11.1875, |
| "frac_dead": 5.809466165374033e-05, |
| "frac_sat": 1.8543667579251633e-07 |
| }, |
| "enc_downs.2.op": { |
| "mean": 0.23879824578762054, |
| "std": 2.575745105743408, |
| "abs_max": 22.625, |
| "frac_dead": 9.43077975534834e-06, |
| "frac_sat": 0.0032885868567973375 |
| }, |
| "enc_blocks.3.norm1": { |
| "mean": 0.0005088050966151059, |
| "std": 1.0007959604263306, |
| "abs_max": 8.825421333312988, |
| "frac_dead": 1.8013848830378265e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.conv1": { |
| "mean": -0.1672450304031372, |
| "std": 1.5013314485549927, |
| "abs_max": 22.5, |
| "frac_dead": 1.896752291941084e-05, |
| "frac_sat": 0.00022898780298419297 |
| }, |
| "enc_blocks.3.t_proj": { |
| "mean": -0.0006834938540123403, |
| "std": 0.057969387620687485, |
| "abs_max": 0.2578125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.norm2": { |
| "mean": -0.002198457717895508, |
| "std": 0.997900128364563, |
| "abs_max": 12.679757118225098, |
| "frac_dead": 5.298190899338806e-07, |
| "frac_sat": 3.0729506761417724e-06 |
| }, |
| "enc_blocks.3.conv2": { |
| "mean": -0.07520721107721329, |
| "std": 1.4643439054489136, |
| "abs_max": 17.75, |
| "frac_dead": 2.0451016098377295e-05, |
| "frac_sat": 0.00010522206866880879 |
| }, |
| "enc_attns.3.norm": { |
| "mean": 0.00023528767633251846, |
| "std": 0.999700129032135, |
| "abs_max": 9.543044090270996, |
| "frac_dead": 1.059638137235197e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.3.qkv": { |
| "mean": -0.003681996837258339, |
| "std": 0.7336653470993042, |
| "abs_max": 11.8125, |
| "frac_dead": 9.278897778131068e-05, |
| "frac_sat": 1.7660636331129353e-07 |
| }, |
| "enc_attns.3.proj": { |
| "mean": -0.004405897576361895, |
| "std": 0.3329181671142578, |
| "abs_max": 2.453125, |
| "frac_dead": 0.00024106768250931054, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.3.op": { |
| "mean": 0.5550792813301086, |
| "std": 3.9987339973449707, |
| "abs_max": 30.375, |
| "frac_dead": 1.0172526344831567e-05, |
| "frac_sat": 0.02196587435901165 |
| }, |
| "enc_blocks.4.norm1": { |
| "mean": -0.00024249155831057578, |
| "std": 0.9996702671051025, |
| "abs_max": 8.92320728302002, |
| "frac_dead": 1.271565793103946e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.conv1": { |
| "mean": -0.22923271358013153, |
| "std": 1.21475088596344, |
| "abs_max": 14.1875, |
| "frac_dead": 2.3100112230167724e-05, |
| "frac_sat": 6.167093670228496e-05 |
| }, |
| "enc_blocks.4.t_proj": { |
| "mean": -0.0025398412253707647, |
| "std": 0.058452896773815155, |
| "abs_max": 0.322265625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.norm2": { |
| "mean": -0.0034613648895174265, |
| "std": 0.9967631101608276, |
| "abs_max": 10.875402450561523, |
| "frac_dead": 1.271565793103946e-06, |
| "frac_sat": 2.755059085757239e-06 |
| }, |
| "enc_blocks.4.conv2": { |
| "mean": 0.05035363510251045, |
| "std": 1.1800023317337036, |
| "abs_max": 8.6875, |
| "frac_dead": 1.4623005881730933e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.skip": { |
| "mean": 0.18852955102920532, |
| "std": 2.690230131149292, |
| "abs_max": 21.875, |
| "frac_dead": 2.7338663130649365e-05, |
| "frac_sat": 0.0027665032539516687 |
| }, |
| "enc_attns.4.norm": { |
| "mean": 5.994405910314526e-06, |
| "std": 0.9989033341407776, |
| "abs_max": 9.629432678222656, |
| "frac_dead": 1.4834934063401306e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.qkv": { |
| "mean": -0.003762299893423915, |
| "std": 0.8487932682037354, |
| "abs_max": 9.25, |
| "frac_dead": 6.668655987596139e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.proj": { |
| "mean": 0.00023298907035496086, |
| "std": 0.16651789844036102, |
| "abs_max": 1.375, |
| "frac_dead": 0.0002608829236123711, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.4.op": { |
| "mean": -0.052241500467061996, |
| "std": 4.246338367462158, |
| "abs_max": 30.375, |
| "frac_dead": 5.086263172415784e-06, |
| "frac_sat": 0.03183576837182045 |
| }, |
| "enc_blocks.5.norm1": { |
| "mean": 0.0003438827989157289, |
| "std": 1.0007237195968628, |
| "abs_max": 10.009206771850586, |
| "frac_dead": 0.0, |
| "frac_sat": 8.477105097881577e-07 |
| }, |
| "enc_blocks.5.conv1": { |
| "mean": -0.3997481167316437, |
| "std": 1.9342615604400635, |
| "abs_max": 14.6875, |
| "frac_dead": 1.0172526344831567e-05, |
| "frac_sat": 0.00036282010842114687 |
| }, |
| "enc_blocks.5.t_proj": { |
| "mean": -0.002505121985450387, |
| "std": 0.057327546179294586, |
| "abs_max": 0.333984375, |
| "frac_dead": 0.00010850694525288418, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.norm2": { |
| "mean": -0.0028387042693793774, |
| "std": 0.9975071549415588, |
| "abs_max": 7.06866455078125, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv2": { |
| "mean": 0.09301067143678665, |
| "std": 1.1804240942001343, |
| "abs_max": 6.21875, |
| "frac_dead": 1.610649997019209e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.norm": { |
| "mean": -4.0283219277625903e-05, |
| "std": 0.9991267919540405, |
| "abs_max": 10.267005920410156, |
| "frac_dead": 2.543131586207892e-06, |
| "frac_sat": 1.6954210195763153e-06 |
| }, |
| "enc_attns.5.qkv": { |
| "mean": -0.004235548432916403, |
| "std": 0.7168772220611572, |
| "abs_max": 6.96875, |
| "frac_dead": 7.62939453125e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.proj": { |
| "mean": 0.003645795863121748, |
| "std": 0.28655537962913513, |
| "abs_max": 1.3828125, |
| "frac_dead": 0.00013648139429278672, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm1": { |
| "mean": 0.00039271556306630373, |
| "std": 1.0009983777999878, |
| "abs_max": 10.273219108581543, |
| "frac_dead": 0.0, |
| "frac_sat": 1.6954210195763153e-06 |
| }, |
| "mid1.conv1": { |
| "mean": -0.6475042700767517, |
| "std": 1.9125096797943115, |
| "abs_max": 15.5, |
| "frac_dead": 8.47710543894209e-06, |
| "frac_sat": 0.00037807889748364687 |
| }, |
| "mid1.t_proj": { |
| "mean": 0.0028974239248782396, |
| "std": 0.05917719379067421, |
| "abs_max": 0.2890625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm2": { |
| "mean": -0.002513372339308262, |
| "std": 0.9980164170265198, |
| "abs_max": 8.224201202392578, |
| "frac_dead": 2.543131586207892e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv2": { |
| "mean": 0.044998347759246826, |
| "std": 1.407365322113037, |
| "abs_max": 8.0, |
| "frac_dead": 1.1867947250721045e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.norm": { |
| "mean": -3.3670828997856006e-05, |
| "std": 0.9990683794021606, |
| "abs_max": 8.834624290466309, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.qkv": { |
| "mean": 0.013007380068302155, |
| "std": 0.6966890692710876, |
| "abs_max": 7.1875, |
| "frac_dead": 7.007740350672975e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.proj": { |
| "mean": 0.0007060468196868896, |
| "std": 0.3016098439693451, |
| "abs_max": 1.3125, |
| "frac_dead": 0.00014326308155432343, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm1": { |
| "mean": 0.0003113076090812683, |
| "std": 1.0007456541061401, |
| "abs_max": 8.782316207885742, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv1": { |
| "mean": -0.5544661283493042, |
| "std": 1.6419425010681152, |
| "abs_max": 14.875, |
| "frac_dead": 1.610649997019209e-05, |
| "frac_sat": 0.00017886691784951836 |
| }, |
| "mid2.t_proj": { |
| "mean": 0.0005331606371328235, |
| "std": 0.059308089315891266, |
| "abs_max": 0.341796875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm2": { |
| "mean": -0.003296496579423547, |
| "std": 0.9975892305374146, |
| "abs_max": 8.295001983642578, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv2": { |
| "mean": 0.06585688143968582, |
| "std": 1.362164855003357, |
| "abs_max": 8.1875, |
| "frac_dead": 1.4411079064302612e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm1": { |
| "mean": 0.00014007091522216797, |
| "std": 1.0002217292785645, |
| "abs_max": 10.676835060119629, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 1.6954210195763153e-06 |
| }, |
| "dec_blocks.0.conv1": { |
| "mean": -1.6476129293441772, |
| "std": 2.691973924636841, |
| "abs_max": 32.25, |
| "frac_dead": 1.1867947250721045e-05, |
| "frac_sat": 0.02121988870203495 |
| }, |
| "dec_blocks.0.t_proj": { |
| "mean": -0.001999007537961006, |
| "std": 0.06088656559586525, |
| "abs_max": 0.2734375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm2": { |
| "mean": -0.00483663147315383, |
| "std": 0.9973129630088806, |
| "abs_max": 10.574625968933105, |
| "frac_dead": 0.0, |
| "frac_sat": 4.238552719471045e-06 |
| }, |
| "dec_blocks.0.conv2": { |
| "mean": -0.00604268629103899, |
| "std": 0.5441529154777527, |
| "abs_max": 2.578125, |
| "frac_dead": 3.560383993317373e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.skip": { |
| "mean": -0.032212819904088974, |
| "std": 1.982886552810669, |
| "abs_max": 13.25, |
| "frac_dead": 2.1192763597355224e-05, |
| "frac_sat": 0.00012122260523028672 |
| }, |
| "dec_attns.0.norm": { |
| "mean": -0.00016459905600640923, |
| "std": 0.9984184503555298, |
| "abs_max": 8.354228973388672, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.qkv": { |
| "mean": -0.009556476026773453, |
| "std": 0.5968071222305298, |
| "abs_max": 7.625, |
| "frac_dead": 8.67490452947095e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.proj": { |
| "mean": 0.008083020336925983, |
| "std": 0.35218313336372375, |
| "abs_max": 1.3125, |
| "frac_dead": 0.0001237657415913418, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.0.op": { |
| "mean": 0.3276306092739105, |
| "std": 2.0408554077148438, |
| "abs_max": 14.5625, |
| "frac_dead": 9.960598617908545e-06, |
| "frac_sat": 0.0007154676713980734 |
| }, |
| "dec_blocks.1.norm1": { |
| "mean": -8.537082612747326e-05, |
| "std": 1.0001024007797241, |
| "abs_max": 10.442710876464844, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 3.178914482759865e-07 |
| }, |
| "dec_blocks.1.conv1": { |
| "mean": -1.339745283126831, |
| "std": 2.232980728149414, |
| "abs_max": 34.25, |
| "frac_dead": 1.038445407175459e-05, |
| "frac_sat": 0.00516870291903615 |
| }, |
| "dec_blocks.1.t_proj": { |
| "mean": -0.002125428058207035, |
| "std": 0.05761529505252838, |
| "abs_max": 0.34375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.norm2": { |
| "mean": -0.004831163212656975, |
| "std": 0.995678186416626, |
| "abs_max": 12.500800132751465, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 6.993611805228284e-06 |
| }, |
| "dec_blocks.1.conv2": { |
| "mean": 0.0020734877325594425, |
| "std": 0.42178505659103394, |
| "abs_max": 1.859375, |
| "frac_dead": 3.24249267578125e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.skip": { |
| "mean": -0.018268505111336708, |
| "std": 1.4295871257781982, |
| "abs_max": 9.125, |
| "frac_dead": 2.5219387680408545e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.norm": { |
| "mean": 0.00011205342161701992, |
| "std": 0.9994926452636719, |
| "abs_max": 8.612427711486816, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.qkv": { |
| "mean": 0.01128898374736309, |
| "std": 0.6574664115905762, |
| "abs_max": 6.59375, |
| "frac_dead": 7.043061486911029e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.proj": { |
| "mean": 0.0020239518489688635, |
| "std": 0.3653465509414673, |
| "abs_max": 1.3984375, |
| "frac_dead": 0.000118255615234375, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.1.op": { |
| "mean": 0.015648458153009415, |
| "std": 1.215781569480896, |
| "abs_max": 8.9375, |
| "frac_dead": 1.340442213404458e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm1": { |
| "mean": 3.191718496964313e-05, |
| "std": 0.9998952150344849, |
| "abs_max": 9.021163940429688, |
| "frac_dead": 3.8853397654747823e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.conv1": { |
| "mean": -0.6205039620399475, |
| "std": 2.467486619949341, |
| "abs_max": 38.75, |
| "frac_dead": 7.417466804326978e-06, |
| "frac_sat": 0.006686422508209944 |
| }, |
| "dec_blocks.2.t_proj": { |
| "mean": -0.005355479661375284, |
| "std": 0.060733918100595474, |
| "abs_max": 0.271484375, |
| "frac_dead": 0.00010850694525288418, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm2": { |
| "mean": -0.005140045657753944, |
| "std": 0.9974221587181091, |
| "abs_max": 12.372010231018066, |
| "frac_dead": 3.178914482759865e-07, |
| "frac_sat": 1.600053656147793e-05 |
| }, |
| "dec_blocks.2.conv2": { |
| "mean": -0.0005674130516126752, |
| "std": 0.5642880797386169, |
| "abs_max": 3.90625, |
| "frac_dead": 2.564324313425459e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.skip": { |
| "mean": 0.0454326756298542, |
| "std": 1.3182048797607422, |
| "abs_max": 10.125, |
| "frac_dead": 3.0517578125e-05, |
| "frac_sat": 1.059638137235197e-07 |
| }, |
| "dec_attns.2.norm": { |
| "mean": 2.561923611210659e-05, |
| "std": 0.9981249570846558, |
| "abs_max": 8.257494926452637, |
| "frac_dead": 5.298190899338806e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.qkv": { |
| "mean": 0.015855487436056137, |
| "std": 0.5967628955841064, |
| "abs_max": 7.1875, |
| "frac_dead": 0.00010296150867361575, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.proj": { |
| "mean": -0.0011597300181165338, |
| "std": 0.3089199960231781, |
| "abs_max": 1.1796875, |
| "frac_dead": 0.00023799472546670586, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.2.op": { |
| "mean": 0.05098172277212143, |
| "std": 1.2449263334274292, |
| "abs_max": 7.40625, |
| "frac_dead": 1.7510519683128223e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm1": { |
| "mean": 0.0001633289357414469, |
| "std": 1.0001723766326904, |
| "abs_max": 7.487753391265869, |
| "frac_dead": 9.5367431640625e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.conv1": { |
| "mean": -0.9776155948638916, |
| "std": 1.510038137435913, |
| "abs_max": 18.375, |
| "frac_dead": 1.1258654922130518e-05, |
| "frac_sat": 0.00020133124780841172 |
| }, |
| "dec_blocks.3.t_proj": { |
| "mean": -0.00041362602496519685, |
| "std": 0.06447338312864304, |
| "abs_max": 0.306640625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm2": { |
| "mean": -0.004704537335783243, |
| "std": 0.9959903955459595, |
| "abs_max": 11.89158821105957, |
| "frac_dead": 7.417467031700653e-07, |
| "frac_sat": 7.417467031700653e-07 |
| }, |
| "dec_blocks.3.conv2": { |
| "mean": 0.05290665104985237, |
| "std": 0.4503594934940338, |
| "abs_max": 3.375, |
| "frac_dead": 4.665056985686533e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.skip": { |
| "mean": 0.03325294330716133, |
| "std": 1.0477484464645386, |
| "abs_max": 7.46875, |
| "frac_dead": 4.121992242289707e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.3.op": { |
| "mean": 0.03040865808725357, |
| "std": 1.2022557258605957, |
| "abs_max": 7.25, |
| "frac_dead": 1.8397966414340772e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm1": { |
| "mean": 0.00019849088857881725, |
| "std": 1.0003993511199951, |
| "abs_max": 7.175535202026367, |
| "frac_dead": 7.329163622671331e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.conv1": { |
| "mean": -0.652289092540741, |
| "std": 1.7954871654510498, |
| "abs_max": 9.375, |
| "frac_dead": 1.0304980605724268e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.t_proj": { |
| "mean": -0.00229295389726758, |
| "std": 0.06348087638616562, |
| "abs_max": 0.29296875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm2": { |
| "mean": -0.0027420183178037405, |
| "std": 0.9977593421936035, |
| "abs_max": 4.8135457038879395, |
| "frac_dead": 1.086129145733139e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.conv2": { |
| "mean": -0.09926741570234299, |
| "std": 0.6462329030036926, |
| "abs_max": 3.515625, |
| "frac_dead": 4.35378824477084e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.skip": { |
| "mean": -0.07995840162038803, |
| "std": 1.1788166761398315, |
| "abs_max": 7.59375, |
| "frac_dead": 4.5286284148460254e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.4.op": { |
| "mean": -0.2425907403230667, |
| "std": 1.8716286420822144, |
| "abs_max": 14.5625, |
| "frac_dead": 1.8404589354759082e-05, |
| "frac_sat": 0.00025292899226769805 |
| }, |
| "dec_blocks.5.norm1": { |
| "mean": -0.0022172522731125355, |
| "std": 0.9993937015533447, |
| "abs_max": 8.945484161376953, |
| "frac_dead": 8.245309004450974e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.conv1": { |
| "mean": -0.19136612117290497, |
| "std": 1.2564855813980103, |
| "abs_max": 9.3125, |
| "frac_dead": 1.5394554793601856e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.t_proj": { |
| "mean": 0.002546150702983141, |
| "std": 0.09909532964229584, |
| "abs_max": 0.60546875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.norm2": { |
| "mean": 0.00035895511973649263, |
| "std": 1.0026216506958008, |
| "abs_max": 10.721990585327148, |
| "frac_dead": 1.6225709487116546e-07, |
| "frac_sat": 4.967053612858763e-08 |
| }, |
| "dec_blocks.5.conv2": { |
| "mean": -0.31369104981422424, |
| "std": 1.3686388731002808, |
| "abs_max": 13.8125, |
| "frac_dead": 2.8093656510463916e-05, |
| "frac_sat": 1.231829355674563e-05 |
| }, |
| "dec_blocks.5.skip": { |
| "mean": -0.03304848447442055, |
| "std": 1.2728080749511719, |
| "abs_max": 8.125, |
| "frac_dead": 4.986591011402197e-05, |
| "frac_sat": 0.0 |
| }, |
| "out_norm": { |
| "mean": -0.0035718195140361786, |
| "std": 0.9979620575904846, |
| "abs_max": 8.589007377624512, |
| "frac_dead": 3.1458006333195954e-07, |
| "frac_sat": 0.0 |
| }, |
| "out_conv": { |
| "mean": 0.14288969337940216, |
| "std": 0.2213430255651474, |
| "abs_max": 0.91796875, |
| "frac_dead": 1.3139512702764478e-05, |
| "frac_sat": 0.0 |
| } |
| } |
| } |
