| { |
| "iter": 500, |
| "per_layer": { |
| "time_mlp.1.weight": { |
| "w_norm": 18.595300674438477, |
| "g_norm": 0.0813865214586258, |
| "shape": [ |
| 1024, |
| 256 |
| ] |
| }, |
| "time_mlp.1.bias": { |
| "w_norm": 1.1589714288711548, |
| "g_norm": 0.007196313701570034, |
| "shape": [ |
| 1024 |
| ] |
| }, |
| "time_mlp.3.weight": { |
| "w_norm": 9.38232135772705, |
| "g_norm": 0.038036804646253586, |
| "shape": [ |
| 256, |
| 1024 |
| ] |
| }, |
| "time_mlp.3.bias": { |
| "w_norm": 0.29130446910858154, |
| "g_norm": 0.0083179771900177, |
| "shape": [ |
| 256 |
| ] |
| }, |
| "input_conv.weight": { |
| "w_norm": 7.941638469696045, |
| "g_norm": 0.07811103761196136, |
| "shape": [ |
| 192, |
| 6, |
| 3, |
| 3 |
| ] |
| }, |
| "input_conv.bias": { |
| "w_norm": 1.1801764965057373, |
| "g_norm": 0.05344012752175331, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.weight": { |
| "w_norm": 13.835872650146484, |
| "g_norm": 0.015104694291949272, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.bias": { |
| "w_norm": 0.03625425696372986, |
| "g_norm": 0.022762326523661613, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv1.weight": { |
| "w_norm": 8.11551570892334, |
| "g_norm": 0.1937582790851593, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv1.bias": { |
| "w_norm": 0.2021755576133728, |
| "g_norm": 0.019418831914663315, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.t_proj.weight": { |
| "w_norm": 8.000494956970215, |
| "g_norm": 0.014280546456575394, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.0.t_proj.bias": { |
| "w_norm": 0.5028907656669617, |
| "g_norm": 0.01941794902086258, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.weight": { |
| "w_norm": 13.850933074951172, |
| "g_norm": 0.00578329199925065, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.bias": { |
| "w_norm": 0.019755441695451736, |
| "g_norm": 0.007538977544754744, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv2.weight": { |
| "w_norm": 8.096028327941895, |
| "g_norm": 0.06561890244483948, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv2.bias": { |
| "w_norm": 0.187120258808136, |
| "g_norm": 0.0059786029160022736, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.weight": { |
| "w_norm": 13.851934432983398, |
| "g_norm": 9.219303319696337e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.bias": { |
| "w_norm": 0.015200300142168999, |
| "g_norm": 0.00011359907512087375, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv1.weight": { |
| "w_norm": 8.089102745056152, |
| "g_norm": 0.0032792543061077595, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv1.bias": { |
| "w_norm": 0.19240157306194305, |
| "g_norm": 0.000176840418134816, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.t_proj.weight": { |
| "w_norm": 7.998848915100098, |
| "g_norm": 0.0001282700541196391, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.1.t_proj.bias": { |
| "w_norm": 0.5218672752380371, |
| "g_norm": 0.00017683516489341855, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.weight": { |
| "w_norm": 13.833203315734863, |
| "g_norm": 0.00020846868574153632, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.bias": { |
| "w_norm": 0.02090640738606453, |
| "g_norm": 0.00021477529662661254, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv2.weight": { |
| "w_norm": 8.047603607177734, |
| "g_norm": 0.006150349508970976, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv2.bias": { |
| "w_norm": 0.2068602740764618, |
| "g_norm": 0.0003937286091968417, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.weight": { |
| "w_norm": 13.853577613830566, |
| "g_norm": 2.653978117450606e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.bias": { |
| "w_norm": 0.020222244784235954, |
| "g_norm": 3.004993050126359e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.conv1.weight": { |
| "w_norm": 11.423590660095215, |
| "g_norm": 0.0007072354201227427, |
| "shape": [ |
| 384, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv1.bias": { |
| "w_norm": 0.2670600712299347, |
| "g_norm": 3.2856267353054136e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.t_proj.weight": { |
| "w_norm": 11.348468780517578, |
| "g_norm": 2.6057350623887032e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.2.t_proj.bias": { |
| "w_norm": 0.7200393080711365, |
| "g_norm": 3.286291394033469e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.weight": { |
| "w_norm": 19.526947021484375, |
| "g_norm": 2.8658503651968203e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.bias": { |
| "w_norm": 0.052438266575336456, |
| "g_norm": 3.415972969378345e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.conv2.weight": { |
| "w_norm": 11.376154899597168, |
| "g_norm": 0.0018718303181231022, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv2.bias": { |
| "w_norm": 0.20025207102298737, |
| "g_norm": 7.625852595083416e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.skip.weight": { |
| "w_norm": 11.280326843261719, |
| "g_norm": 0.0024632003623992205, |
| "shape": [ |
| 384, |
| 192, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.2.skip.bias": { |
| "w_norm": 0.8246392607688904, |
| "g_norm": 7.625852595083416e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.weight": { |
| "w_norm": 19.596302032470703, |
| "g_norm": 6.7835412664862815e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.bias": { |
| "w_norm": 0.026706550270318985, |
| "g_norm": 6.360353381751338e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv1.weight": { |
| "w_norm": 11.475268363952637, |
| "g_norm": 0.00031782532460056245, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv1.bias": { |
| "w_norm": 0.1944141387939453, |
| "g_norm": 9.2924574346398e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.t_proj.weight": { |
| "w_norm": 11.324196815490723, |
| "g_norm": 7.110313617886277e-06, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.3.t_proj.bias": { |
| "w_norm": 0.691976010799408, |
| "g_norm": 9.291392416344024e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.weight": { |
| "w_norm": 19.542306900024414, |
| "g_norm": 2.2274725779425353e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.bias": { |
| "w_norm": 0.04708161577582359, |
| "g_norm": 1.5845758753130212e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv2.weight": { |
| "w_norm": 11.365864753723145, |
| "g_norm": 0.0007113237516023219, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv2.bias": { |
| "w_norm": 0.1897016316652298, |
| "g_norm": 2.332814437977504e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.weight": { |
| "w_norm": 19.591983795166016, |
| "g_norm": 1.2785332728526555e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.bias": { |
| "w_norm": 0.032775815576314926, |
| "g_norm": 1.206148044730071e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.conv1.weight": { |
| "w_norm": 16.21708869934082, |
| "g_norm": 0.00042211293475702405, |
| "shape": [ |
| 768, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv1.bias": { |
| "w_norm": 0.279443621635437, |
| "g_norm": 1.3359977856453042e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.t_proj.weight": { |
| "w_norm": 16.025165557861328, |
| "g_norm": 1.004903060675133e-05, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.4.t_proj.bias": { |
| "w_norm": 0.9901605248451233, |
| "g_norm": 1.3356218914850615e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.weight": { |
| "w_norm": 27.63109588623047, |
| "g_norm": 1.321973559242906e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.bias": { |
| "w_norm": 0.07533392310142517, |
| "g_norm": 1.3589736227004323e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.conv2.weight": { |
| "w_norm": 16.13887596130371, |
| "g_norm": 0.0006326561560854316, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv2.bias": { |
| "w_norm": 0.19331632554531097, |
| "g_norm": 1.6220848920056596e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.skip.weight": { |
| "w_norm": 15.950276374816895, |
| "g_norm": 0.0005513212527148426, |
| "shape": [ |
| 768, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.4.skip.bias": { |
| "w_norm": 0.8089227676391602, |
| "g_norm": 1.6220848920056596e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.weight": { |
| "w_norm": 27.711660385131836, |
| "g_norm": 7.78115281718783e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.bias": { |
| "w_norm": 0.04768697917461395, |
| "g_norm": 6.8741815084649716e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv1.weight": { |
| "w_norm": 16.37407684326172, |
| "g_norm": 0.00023456118651665747, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv1.bias": { |
| "w_norm": 0.2019558846950531, |
| "g_norm": 4.756613179779379e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.t_proj.weight": { |
| "w_norm": 15.995716094970703, |
| "g_norm": 3.4388413041597232e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.5.t_proj.bias": { |
| "w_norm": 0.9961099028587341, |
| "g_norm": 4.755159352498595e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.weight": { |
| "w_norm": 27.65757942199707, |
| "g_norm": 1.1370511856512167e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.bias": { |
| "w_norm": 0.06127138435840607, |
| "g_norm": 9.395703273185063e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv2.weight": { |
| "w_norm": 16.235984802246094, |
| "g_norm": 0.0004213395877741277, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv2.bias": { |
| "w_norm": 0.19209671020507812, |
| "g_norm": 1.1171143341925927e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.3.norm.weight": { |
| "w_norm": 19.575828552246094, |
| "g_norm": 5.652593245031312e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.norm.bias": { |
| "w_norm": 0.02052447572350502, |
| "g_norm": 7.894850568845868e-06, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.qkv.weight": { |
| "w_norm": 19.62198257446289, |
| "g_norm": 0.0001925578253576532, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.qkv.bias": { |
| "w_norm": 1.0065840482711792, |
| "g_norm": 1.3265251254779287e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "enc_attns.3.proj.weight": { |
| "w_norm": 11.29690170288086, |
| "g_norm": 0.0001960561639862135, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.proj.bias": { |
| "w_norm": 0.5664379000663757, |
| "g_norm": 2.299301377206575e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.4.norm.weight": { |
| "w_norm": 27.69475555419922, |
| "g_norm": 4.17779165218235e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.norm.bias": { |
| "w_norm": 0.02993551641702652, |
| "g_norm": 6.423387731047114e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.qkv.weight": { |
| "w_norm": 27.85542106628418, |
| "g_norm": 0.0001815570140024647, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.qkv.bias": { |
| "w_norm": 1.0133144855499268, |
| "g_norm": 9.848357876762748e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.4.proj.weight": { |
| "w_norm": 15.99245548248291, |
| "g_norm": 0.00018240074859932065, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.proj.bias": { |
| "w_norm": 0.5735777020454407, |
| "g_norm": 1.5512972822762094e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.weight": { |
| "w_norm": 27.691572189331055, |
| "g_norm": 3.4669185424718307e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.bias": { |
| "w_norm": 0.028933243826031685, |
| "g_norm": 4.15879048887291e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.qkv.weight": { |
| "w_norm": 27.788082122802734, |
| "g_norm": 0.00015120556054171175, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.qkv.bias": { |
| "w_norm": 0.9982998967170715, |
| "g_norm": 6.607310751860496e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.5.proj.weight": { |
| "w_norm": 16.01085662841797, |
| "g_norm": 0.00014587878831662238, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.proj.bias": { |
| "w_norm": 0.596544086933136, |
| "g_norm": 1.08937556433375e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_downs.0.op.weight": { |
| "w_norm": 8.090078353881836, |
| "g_norm": 0.007355941459536552, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.0.op.bias": { |
| "w_norm": 0.18642111122608185, |
| "g_norm": 0.000441454176325351, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.1.op.weight": { |
| "w_norm": 8.03896713256836, |
| "g_norm": 0.0029663110617548227, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.1.op.bias": { |
| "w_norm": 0.1942720115184784, |
| "g_norm": 4.368715963209979e-05, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.2.op.weight": { |
| "w_norm": 11.369796752929688, |
| "g_norm": 0.002160790842026472, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.2.op.bias": { |
| "w_norm": 0.18874460458755493, |
| "g_norm": 2.3616021280759014e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.3.op.weight": { |
| "w_norm": 11.359977722167969, |
| "g_norm": 0.0012149059912189841, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.3.op.bias": { |
| "w_norm": 0.1921733319759369, |
| "g_norm": 1.1493058082123753e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.4.op.weight": { |
| "w_norm": 16.15582275390625, |
| "g_norm": 0.0011253517586737871, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.4.op.bias": { |
| "w_norm": 0.19734814763069153, |
| "g_norm": 1.0824763194250409e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.weight": { |
| "w_norm": 27.711822509765625, |
| "g_norm": 1.0373867553425953e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.bias": { |
| "w_norm": 0.04126482829451561, |
| "g_norm": 6.879010015836684e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv1.weight": { |
| "w_norm": 16.37283706665039, |
| "g_norm": 0.0003461484448052943, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv1.bias": { |
| "w_norm": 0.20190510153770447, |
| "g_norm": 6.445853614422958e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.t_proj.weight": { |
| "w_norm": 16.00591278076172, |
| "g_norm": 4.6713685151189566e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid1.t_proj.bias": { |
| "w_norm": 1.031727910041809, |
| "g_norm": 6.443968686653534e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.weight": { |
| "w_norm": 27.664451599121094, |
| "g_norm": 9.174211299978197e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.bias": { |
| "w_norm": 0.061578843742609024, |
| "g_norm": 6.349394425342325e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv2.weight": { |
| "w_norm": 16.23827362060547, |
| "g_norm": 0.0003607813559938222, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv2.bias": { |
| "w_norm": 0.19554272294044495, |
| "g_norm": 8.114502634271048e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.weight": { |
| "w_norm": 27.695194244384766, |
| "g_norm": 2.296697402925929e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.bias": { |
| "w_norm": 0.029120877385139465, |
| "g_norm": 3.081472186750034e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.qkv.weight": { |
| "w_norm": 27.785205841064453, |
| "g_norm": 0.00010322411253582686, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.qkv.bias": { |
| "w_norm": 0.9866750240325928, |
| "g_norm": 4.757572241942398e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "mid_attn.proj.weight": { |
| "w_norm": 16.010156631469727, |
| "g_norm": 0.00010198546078754589, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.proj.bias": { |
| "w_norm": 0.5767161846160889, |
| "g_norm": 7.791844836901873e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.weight": { |
| "w_norm": 27.712369918823242, |
| "g_norm": 1.7629792637308128e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.bias": { |
| "w_norm": 0.041550856083631516, |
| "g_norm": 1.2748863809974864e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv1.weight": { |
| "w_norm": 16.345983505249023, |
| "g_norm": 0.0003871308290399611, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv1.bias": { |
| "w_norm": 0.19644778966903687, |
| "g_norm": 7.461239420081256e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.t_proj.weight": { |
| "w_norm": 15.993643760681152, |
| "g_norm": 5.395043444877956e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid2.t_proj.bias": { |
| "w_norm": 0.9956456422805786, |
| "g_norm": 7.462194844265468e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.weight": { |
| "w_norm": 27.654315948486328, |
| "g_norm": 9.289041372539941e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.bias": { |
| "w_norm": 0.06868414580821991, |
| "g_norm": 7.056999493215699e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv2.weight": { |
| "w_norm": 16.199613571166992, |
| "g_norm": 0.00027969831717200577, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv2.bias": { |
| "w_norm": 0.19018852710723877, |
| "g_norm": 7.015074061200721e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm1.weight": { |
| "w_norm": 39.191375732421875, |
| "g_norm": 1.2209551641717553e-05, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.norm1.bias": { |
| "w_norm": 0.05582411214709282, |
| "g_norm": 8.90219962457195e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.conv1.weight": { |
| "w_norm": 16.71096420288086, |
| "g_norm": 0.0003418675623834133, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv1.bias": { |
| "w_norm": 0.1404905915260315, |
| "g_norm": 5.017229796067113e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.t_proj.weight": { |
| "w_norm": 16.009931564331055, |
| "g_norm": 3.6284368434280623e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.0.t_proj.bias": { |
| "w_norm": 1.0188021659851074, |
| "g_norm": 5.017205239710165e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.weight": { |
| "w_norm": 27.624052047729492, |
| "g_norm": 1.1853226169478148e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.bias": { |
| "w_norm": 0.08831377327442169, |
| "g_norm": 1.0930866665148642e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.conv2.weight": { |
| "w_norm": 16.116975784301758, |
| "g_norm": 0.0005032767076045275, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv2.bias": { |
| "w_norm": 0.19690799713134766, |
| "g_norm": 1.4778754120925441e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.skip.weight": { |
| "w_norm": 15.933785438537598, |
| "g_norm": 0.0012050033546984196, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.0.skip.bias": { |
| "w_norm": 0.41081660985946655, |
| "g_norm": 1.4778754120925441e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm1.weight": { |
| "w_norm": 39.185028076171875, |
| "g_norm": 4.89912235934753e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.norm1.bias": { |
| "w_norm": 0.052894700318574905, |
| "g_norm": 4.081042334291851e-06, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.conv1.weight": { |
| "w_norm": 16.64948272705078, |
| "g_norm": 0.0003468487993814051, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv1.bias": { |
| "w_norm": 0.13766644895076752, |
| "g_norm": 5.367211088014301e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.t_proj.weight": { |
| "w_norm": 16.036720275878906, |
| "g_norm": 4.022125267511001e-06, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.1.t_proj.bias": { |
| "w_norm": 0.9849644899368286, |
| "g_norm": 5.367090125218965e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.weight": { |
| "w_norm": 27.63217544555664, |
| "g_norm": 1.2087589311704505e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.bias": { |
| "w_norm": 0.08338069170713425, |
| "g_norm": 1.302398322877707e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.conv2.weight": { |
| "w_norm": 16.113073348999023, |
| "g_norm": 0.0011885776184499264, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv2.bias": { |
| "w_norm": 0.1954251527786255, |
| "g_norm": 2.834928091033362e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.skip.weight": { |
| "w_norm": 15.929043769836426, |
| "g_norm": 0.0016937535256147385, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.1.skip.bias": { |
| "w_norm": 0.40631330013275146, |
| "g_norm": 2.834928091033362e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.2.norm1.weight": { |
| "w_norm": 33.94264602661133, |
| "g_norm": 2.3586289898958057e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.norm1.bias": { |
| "w_norm": 0.05327305942773819, |
| "g_norm": 1.8042264855466783e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.conv1.weight": { |
| "w_norm": 11.731250762939453, |
| "g_norm": 0.0008712481940165162, |
| "shape": [ |
| 384, |
| 1152, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv1.bias": { |
| "w_norm": 0.11506623029708862, |
| "g_norm": 1.4161601939122193e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.t_proj.weight": { |
| "w_norm": 11.318585395812988, |
| "g_norm": 1.028281531034736e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.2.t_proj.bias": { |
| "w_norm": 0.7085891962051392, |
| "g_norm": 1.4159787497192156e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.weight": { |
| "w_norm": 19.5445556640625, |
| "g_norm": 4.0533999708713964e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.bias": { |
| "w_norm": 0.05140449106693268, |
| "g_norm": 3.8357742596417665e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.conv2.weight": { |
| "w_norm": 11.374390602111816, |
| "g_norm": 0.0014708536909893155, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv2.bias": { |
| "w_norm": 0.18952299654483795, |
| "g_norm": 5.938815957051702e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.skip.weight": { |
| "w_norm": 11.216829299926758, |
| "g_norm": 0.004319904837757349, |
| "shape": [ |
| 384, |
| 1152, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.2.skip.bias": { |
| "w_norm": 0.3324746787548065, |
| "g_norm": 5.938815957051702e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm1.weight": { |
| "w_norm": 27.71185874938965, |
| "g_norm": 3.5906316043110564e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.norm1.bias": { |
| "w_norm": 0.04073373228311539, |
| "g_norm": 3.671555532491766e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.conv1.weight": { |
| "w_norm": 11.635823249816895, |
| "g_norm": 0.0019167193677276373, |
| "shape": [ |
| 384, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv1.bias": { |
| "w_norm": 0.14201989769935608, |
| "g_norm": 4.213491774862632e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.t_proj.weight": { |
| "w_norm": 11.332667350769043, |
| "g_norm": 3.186779940733686e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.3.t_proj.bias": { |
| "w_norm": 0.7082791328430176, |
| "g_norm": 4.213417923892848e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.weight": { |
| "w_norm": 19.534440994262695, |
| "g_norm": 6.097566074458882e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.bias": { |
| "w_norm": 0.048718083649873734, |
| "g_norm": 8.37421539472416e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.conv2.weight": { |
| "w_norm": 11.369616508483887, |
| "g_norm": 0.003669221419841051, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv2.bias": { |
| "w_norm": 0.19281557202339172, |
| "g_norm": 0.00017457322974223644, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.skip.weight": { |
| "w_norm": 11.228267669677734, |
| "g_norm": 0.0061339568346738815, |
| "shape": [ |
| 384, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.3.skip.bias": { |
| "w_norm": 0.4074385464191437, |
| "g_norm": 0.00017457322974223644, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.4.norm1.weight": { |
| "w_norm": 23.997482299804688, |
| "g_norm": 0.00011013081530109048, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.norm1.bias": { |
| "w_norm": 0.043228741735219955, |
| "g_norm": 0.00013270195631776005, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.conv1.weight": { |
| "w_norm": 8.184913635253906, |
| "g_norm": 0.005579778924584389, |
| "shape": [ |
| 192, |
| 576, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv1.bias": { |
| "w_norm": 0.11560454964637756, |
| "g_norm": 0.00015491456724703312, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.t_proj.weight": { |
| "w_norm": 7.999128341674805, |
| "g_norm": 0.00011237112630624324, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.4.t_proj.bias": { |
| "w_norm": 0.4804953634738922, |
| "g_norm": 0.0001549788867123425, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.weight": { |
| "w_norm": 13.831470489501953, |
| "g_norm": 0.00043151574209332466, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.bias": { |
| "w_norm": 0.01964525133371353, |
| "g_norm": 0.00037272763438522816, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.conv2.weight": { |
| "w_norm": 8.019150733947754, |
| "g_norm": 0.011501150205731392, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv2.bias": { |
| "w_norm": 0.19350716471672058, |
| "g_norm": 0.0006961687467992306, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.skip.weight": { |
| "w_norm": 7.952949047088623, |
| "g_norm": 0.01609424687922001, |
| "shape": [ |
| 192, |
| 576, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.4.skip.bias": { |
| "w_norm": 0.34671664237976074, |
| "g_norm": 0.0006961687467992306, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm1.weight": { |
| "w_norm": 19.585325241088867, |
| "g_norm": 0.001735931378789246, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.norm1.bias": { |
| "w_norm": 0.036612581461668015, |
| "g_norm": 0.0026527568697929382, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.conv1.weight": { |
| "w_norm": 8.18410873413086, |
| "g_norm": 0.037934672087430954, |
| "shape": [ |
| 192, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv1.bias": { |
| "w_norm": 0.13837724924087524, |
| "g_norm": 0.0017917185323312879, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.t_proj.weight": { |
| "w_norm": 8.034667015075684, |
| "g_norm": 0.0013399251038208604, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.5.t_proj.bias": { |
| "w_norm": 0.5041279792785645, |
| "g_norm": 0.0017916002543643117, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.weight": { |
| "w_norm": 13.868912696838379, |
| "g_norm": 0.0018698613857850432, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.bias": { |
| "w_norm": 0.022572237998247147, |
| "g_norm": 0.0022469060495495796, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.conv2.weight": { |
| "w_norm": 8.10611343383789, |
| "g_norm": 0.0494694784283638, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv2.bias": { |
| "w_norm": 0.19638051092624664, |
| "g_norm": 0.002709712367504835, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.skip.weight": { |
| "w_norm": 8.038369178771973, |
| "g_norm": 0.05201084911823273, |
| "shape": [ |
| 192, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.5.skip.bias": { |
| "w_norm": 0.41981905698776245, |
| "g_norm": 0.002709712367504835, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_attns.0.norm.weight": { |
| "w_norm": 27.69066047668457, |
| "g_norm": 7.059255040076096e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.norm.bias": { |
| "w_norm": 0.022313563153147697, |
| "g_norm": 7.783974979247432e-06, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.qkv.weight": { |
| "w_norm": 27.7947940826416, |
| "g_norm": 0.0002512983337510377, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.qkv.bias": { |
| "w_norm": 1.0090641975402832, |
| "g_norm": 9.96327435132116e-06, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.0.proj.weight": { |
| "w_norm": 16.015512466430664, |
| "g_norm": 0.00023575192608404905, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.proj.bias": { |
| "w_norm": 0.5766009092330933, |
| "g_norm": 1.5225058632495347e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.weight": { |
| "w_norm": 27.70824432373047, |
| "g_norm": 1.0122570529347286e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.bias": { |
| "w_norm": 0.025859301909804344, |
| "g_norm": 1.2997406884096563e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.qkv.weight": { |
| "w_norm": 27.807323455810547, |
| "g_norm": 0.00034215542837046087, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.qkv.bias": { |
| "w_norm": 1.005326271057129, |
| "g_norm": 1.735401565383654e-05, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.1.proj.weight": { |
| "w_norm": 16.04290771484375, |
| "g_norm": 0.00034307295572943985, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.proj.bias": { |
| "w_norm": 0.5798501372337341, |
| "g_norm": 2.702936762943864e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.2.norm.weight": { |
| "w_norm": 19.579172134399414, |
| "g_norm": 1.5314006304834038e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.norm.bias": { |
| "w_norm": 0.017062561586499214, |
| "g_norm": 2.1284606191329658e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.qkv.weight": { |
| "w_norm": 19.6417293548584, |
| "g_norm": 0.0004970437148585916, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.qkv.bias": { |
| "w_norm": 0.9886179566383362, |
| "g_norm": 3.3770891604945064e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_attns.2.proj.weight": { |
| "w_norm": 11.308408737182617, |
| "g_norm": 0.0004960380611009896, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.proj.bias": { |
| "w_norm": 0.5879140496253967, |
| "g_norm": 5.74711557419505e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.0.op.weight": { |
| "w_norm": 16.0877628326416, |
| "g_norm": 0.001645346055738628, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.0.op.bias": { |
| "w_norm": 0.20079049468040466, |
| "g_norm": 1.3754704013990704e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.1.op.weight": { |
| "w_norm": 16.08888053894043, |
| "g_norm": 0.003802034305408597, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.1.op.bias": { |
| "w_norm": 0.18735860288143158, |
| "g_norm": 3.066011777264066e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.2.op.weight": { |
| "w_norm": 11.317456245422363, |
| "g_norm": 0.008105107583105564, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.2.op.bias": { |
| "w_norm": 0.19506803154945374, |
| "g_norm": 8.368221460841596e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.3.op.weight": { |
| "w_norm": 11.315321922302246, |
| "g_norm": 0.01358941849321127, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.3.op.bias": { |
| "w_norm": 0.1970231831073761, |
| "g_norm": 0.0002463943383190781, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.4.op.weight": { |
| "w_norm": 7.9966630935668945, |
| "g_norm": 0.04155389592051506, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.4.op.bias": { |
| "w_norm": 0.19377826154232025, |
| "g_norm": 0.0008116448298096657, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.weight": { |
| "w_norm": 13.845489501953125, |
| "g_norm": 0.008532080799341202, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.bias": { |
| "w_norm": 0.024909643456339836, |
| "g_norm": 0.010218916460871696, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_conv.weight": { |
| "w_norm": 0.09844083338975906, |
| "g_norm": 0.9382269978523254, |
| "shape": [ |
| 3, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "out_conv.bias": { |
| "w_norm": 0.005248754285275936, |
| "g_norm": 0.058636076748371124, |
| "shape": [ |
| 3 |
| ] |
| } |
| }, |
| "activations": { |
| "input_conv": { |
| "mean": 0.005270535591989756, |
| "std": 0.4314875900745392, |
| "abs_max": 3.140625, |
| "frac_dead": 0.00039295354508794844, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.1": { |
| "mean": 0.020573442801833153, |
| "std": 0.3014470338821411, |
| "abs_max": 2.390625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.3": { |
| "mean": 0.013320415280759335, |
| "std": 0.1021292507648468, |
| "abs_max": 0.46875, |
| "frac_dead": 0.00032552084303461015, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm1": { |
| "mean": -0.001979447901248932, |
| "std": 0.9984237551689148, |
| "abs_max": 6.403844833374023, |
| "frac_dead": 1.238452114193933e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv1": { |
| "mean": -0.007452946621924639, |
| "std": 0.5369975566864014, |
| "abs_max": 4.4375, |
| "frac_dead": 6.309151649475098e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.t_proj": { |
| "mean": -0.0015568019589409232, |
| "std": 0.05406259372830391, |
| "abs_max": 0.2451171875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm2": { |
| "mean": -0.0009606948588043451, |
| "std": 0.9998885989189148, |
| "abs_max": 7.327714920043945, |
| "frac_dead": 6.192260570969665e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv2": { |
| "mean": -0.0457494780421257, |
| "std": 0.7087180614471436, |
| "abs_max": 8.5, |
| "frac_dead": 4.7124096454354e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.0.op": { |
| "mean": 0.1254103183746338, |
| "std": 1.4901121854782104, |
| "abs_max": 17.875, |
| "frac_dead": 2.2225909560802393e-05, |
| "frac_sat": 5.988280099700205e-05 |
| }, |
| "enc_blocks.1.norm1": { |
| "mean": -0.000360555131919682, |
| "std": 1.0000308752059937, |
| "abs_max": 9.678850173950195, |
| "frac_dead": 9.006924415189133e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv1": { |
| "mean": 0.07103142887353897, |
| "std": 0.9594767093658447, |
| "abs_max": 11.5625, |
| "frac_dead": 3.520647805999033e-05, |
| "frac_sat": 1.059638137235197e-07 |
| }, |
| "enc_blocks.1.t_proj": { |
| "mean": 0.0024000913836061954, |
| "std": 0.04997396841645241, |
| "abs_max": 0.1552734375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.norm2": { |
| "mean": -0.0016624480485916138, |
| "std": 0.9981207847595215, |
| "abs_max": 9.288371086120605, |
| "frac_dead": 4.5034622075945663e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.conv2": { |
| "mean": -0.0022635753266513348, |
| "std": 0.6630864143371582, |
| "abs_max": 9.875, |
| "frac_dead": 6.64260660414584e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.1.op": { |
| "mean": -0.0900968611240387, |
| "std": 2.5945587158203125, |
| "abs_max": 25.5, |
| "frac_dead": 1.4781951904296875e-05, |
| "frac_sat": 0.004522058647125959 |
| }, |
| "enc_blocks.2.norm1": { |
| "mean": 0.0002139185817213729, |
| "std": 1.0010061264038086, |
| "abs_max": 9.153732299804688, |
| "frac_dead": 5.298190899338806e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.conv1": { |
| "mean": -0.14553506672382355, |
| "std": 1.095105767250061, |
| "abs_max": 13.0, |
| "frac_dead": 3.62396240234375e-05, |
| "frac_sat": 2.5166405066556763e-06 |
| }, |
| "enc_blocks.2.t_proj": { |
| "mean": -0.002175527159124613, |
| "std": 0.04917168989777565, |
| "abs_max": 0.193359375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.norm2": { |
| "mean": -0.0024002897553145885, |
| "std": 0.996701180934906, |
| "abs_max": 11.8389253616333, |
| "frac_dead": 6.092919306865952e-07, |
| "frac_sat": 1.721911985441693e-06 |
| }, |
| "enc_blocks.2.conv2": { |
| "mean": 0.06730736792087555, |
| "std": 0.6235703229904175, |
| "abs_max": 6.96875, |
| "frac_dead": 3.822644430329092e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.skip": { |
| "mean": 0.08201966434717178, |
| "std": 1.3968431949615479, |
| "abs_max": 17.25, |
| "frac_dead": 9.748670709086582e-05, |
| "frac_sat": 7.735358667559922e-05 |
| }, |
| "enc_downs.2.op": { |
| "mean": 0.2103821188211441, |
| "std": 1.7188763618469238, |
| "abs_max": 16.25, |
| "frac_dead": 1.5788607925060205e-05, |
| "frac_sat": 6.283654511207715e-05 |
| }, |
| "enc_blocks.3.norm1": { |
| "mean": 0.00011012620234396309, |
| "std": 1.0003811120986938, |
| "abs_max": 9.707320213317871, |
| "frac_dead": 5.298190899338806e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.conv1": { |
| "mean": 0.12389453500509262, |
| "std": 1.1216708421707153, |
| "abs_max": 15.4375, |
| "frac_dead": 2.3523965865024365e-05, |
| "frac_sat": 6.357828624459216e-06 |
| }, |
| "enc_blocks.3.t_proj": { |
| "mean": -0.002851994475349784, |
| "std": 0.04747543856501579, |
| "abs_max": 0.1826171875, |
| "frac_dead": 0.00010850694525288418, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.norm2": { |
| "mean": -0.0023243087343871593, |
| "std": 0.9972028732299805, |
| "abs_max": 12.23245620727539, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 1.4834934063401306e-06 |
| }, |
| "enc_blocks.3.conv2": { |
| "mean": 0.09541334211826324, |
| "std": 0.7750728726387024, |
| "abs_max": 11.0625, |
| "frac_dead": 4.132588583161123e-05, |
| "frac_sat": 8.477105097881577e-07 |
| }, |
| "enc_attns.3.norm": { |
| "mean": 3.4679967939155176e-05, |
| "std": 0.999211311340332, |
| "abs_max": 10.729257583618164, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 4.238552548940788e-07 |
| }, |
| "enc_attns.3.qkv": { |
| "mean": -0.011044418439269066, |
| "std": 0.6183685660362244, |
| "abs_max": 8.0625, |
| "frac_dead": 0.00011701937182806432, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.3.proj": { |
| "mean": -0.0013116929912939668, |
| "std": 0.2898203730583191, |
| "abs_max": 2.421875, |
| "frac_dead": 0.00019571516895666718, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.3.op": { |
| "mean": -0.13180769979953766, |
| "std": 1.869809627532959, |
| "abs_max": 18.125, |
| "frac_dead": 9.324815437139478e-06, |
| "frac_sat": 0.00029754638671875 |
| }, |
| "enc_blocks.4.norm1": { |
| "mean": -0.00013007968664169312, |
| "std": 1.0002660751342773, |
| "abs_max": 8.761725425720215, |
| "frac_dead": 1.271565793103946e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.conv1": { |
| "mean": -0.24413247406482697, |
| "std": 0.9627959728240967, |
| "abs_max": 10.0, |
| "frac_dead": 3.221299994038418e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.t_proj": { |
| "mean": -0.00039894311339594424, |
| "std": 0.04767155647277832, |
| "abs_max": 0.21875, |
| "frac_dead": 0.00016276042151730508, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.norm2": { |
| "mean": -0.002712855814024806, |
| "std": 0.9977284669876099, |
| "abs_max": 9.868551254272461, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.conv2": { |
| "mean": 0.0014793144073337317, |
| "std": 0.7221308946609497, |
| "abs_max": 6.0625, |
| "frac_dead": 2.649095404194668e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.skip": { |
| "mean": 0.03583603352308273, |
| "std": 1.0848972797393799, |
| "abs_max": 10.6875, |
| "frac_dead": 8.0108642578125e-05, |
| "frac_sat": 6.35782896551973e-07 |
| }, |
| "enc_attns.4.norm": { |
| "mean": -4.4359101593727246e-05, |
| "std": 0.9994379878044128, |
| "abs_max": 11.126827239990234, |
| "frac_dead": 1.0596381798677612e-06, |
| "frac_sat": 4.238552548940788e-07 |
| }, |
| "enc_attns.4.qkv": { |
| "mean": 0.009110452607274055, |
| "std": 0.6766706705093384, |
| "abs_max": 7.71875, |
| "frac_dead": 7.876643212512136e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.proj": { |
| "mean": -0.011879508383572102, |
| "std": 0.3317599892616272, |
| "abs_max": 2.6875, |
| "frac_dead": 0.00016615126514807343, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.4.op": { |
| "mean": 0.04674041271209717, |
| "std": 2.207994222640991, |
| "abs_max": 18.5, |
| "frac_dead": 6.781684078305261e-06, |
| "frac_sat": 0.00032552084303461015 |
| }, |
| "enc_blocks.5.norm1": { |
| "mean": 6.525673961732537e-05, |
| "std": 1.000154972076416, |
| "abs_max": 6.9245285987854, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv1": { |
| "mean": -0.17496569454669952, |
| "std": 1.6681394577026367, |
| "abs_max": 11.9375, |
| "frac_dead": 1.0172526344831567e-05, |
| "frac_sat": 2.0345052689663135e-05 |
| }, |
| "enc_blocks.5.t_proj": { |
| "mean": -0.0018383245915174484, |
| "std": 0.04849156364798546, |
| "abs_max": 0.2314453125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.norm2": { |
| "mean": -0.0017972373170778155, |
| "std": 0.9986117482185364, |
| "abs_max": 6.892906665802002, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv2": { |
| "mean": 0.12745817005634308, |
| "std": 1.3628087043762207, |
| "abs_max": 6.0625, |
| "frac_dead": 1.695421087788418e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.norm": { |
| "mean": 8.80869702086784e-05, |
| "std": 0.9993440508842468, |
| "abs_max": 6.990722179412842, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.qkv": { |
| "mean": 0.0036823570262640715, |
| "std": 0.6195452809333801, |
| "abs_max": 6.0625, |
| "frac_dead": 8.053249621298164e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.proj": { |
| "mean": 0.0019204264972358942, |
| "std": 0.3217765986919403, |
| "abs_max": 2.390625, |
| "frac_dead": 0.0001313951361225918, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm1": { |
| "mean": 1.6457504443678772e-06, |
| "std": 0.9999829530715942, |
| "abs_max": 6.66753625869751, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv1": { |
| "mean": -0.07720707356929779, |
| "std": 0.9930728673934937, |
| "abs_max": 9.1875, |
| "frac_dead": 1.695421087788418e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid1.t_proj": { |
| "mean": 0.002672894624993205, |
| "std": 0.04993588104844093, |
| "abs_max": 0.1875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm2": { |
| "mean": -0.0021396661177277565, |
| "std": 0.9983307123184204, |
| "abs_max": 8.67036247253418, |
| "frac_dead": 3.3908420391526306e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv2": { |
| "mean": 0.028066406026482582, |
| "std": 0.9323076605796814, |
| "abs_max": 7.0, |
| "frac_dead": 1.52587890625e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.norm": { |
| "mean": -7.023000307526672e-06, |
| "std": 0.9994739890098572, |
| "abs_max": 6.992171287536621, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.qkv": { |
| "mean": 0.002624101471155882, |
| "std": 0.6168164014816284, |
| "abs_max": 4.84375, |
| "frac_dead": 8.307563257403672e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.proj": { |
| "mean": 0.0016497158212587237, |
| "std": 0.28975075483322144, |
| "abs_max": 1.234375, |
| "frac_dead": 0.00014326308155432343, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm1": { |
| "mean": -4.525606709648855e-05, |
| "std": 1.0000933408737183, |
| "abs_max": 7.091708660125732, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv1": { |
| "mean": -0.1754794418811798, |
| "std": 0.8759015202522278, |
| "abs_max": 6.1875, |
| "frac_dead": 2.5431314497836865e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid2.t_proj": { |
| "mean": -0.00023175828391686082, |
| "std": 0.04861638322472572, |
| "abs_max": 0.2177734375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm2": { |
| "mean": -0.002224498428404331, |
| "std": 0.9979929327964783, |
| "abs_max": 6.991872787475586, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv2": { |
| "mean": 0.04473405331373215, |
| "std": 1.2036309242248535, |
| "abs_max": 8.0, |
| "frac_dead": 1.8649630874278955e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm1": { |
| "mean": 0.00011099460971308872, |
| "std": 1.00041663646698, |
| "abs_max": 7.442845821380615, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.conv1": { |
| "mean": -0.870387077331543, |
| "std": 1.9311200380325317, |
| "abs_max": 17.25, |
| "frac_dead": 5.086263172415784e-06, |
| "frac_sat": 0.0007129245204851031 |
| }, |
| "dec_blocks.0.t_proj": { |
| "mean": -0.0026858618948608637, |
| "std": 0.04970140382647514, |
| "abs_max": 0.201171875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm2": { |
| "mean": -0.003192740259692073, |
| "std": 0.9973447322845459, |
| "abs_max": 7.285301208496094, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.conv2": { |
| "mean": -0.059047628194093704, |
| "std": 0.7799074649810791, |
| "abs_max": 6.0625, |
| "frac_dead": 1.7801919966586865e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.skip": { |
| "mean": -0.04623860865831375, |
| "std": 1.4913451671600342, |
| "abs_max": 9.75, |
| "frac_dead": 2.1192763597355224e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.norm": { |
| "mean": -5.304647856974043e-05, |
| "std": 0.9991397857666016, |
| "abs_max": 6.559094429016113, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.qkv": { |
| "mean": -0.005534946918487549, |
| "std": 0.6146882176399231, |
| "abs_max": 4.3125, |
| "frac_dead": 8.533619256922975e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.proj": { |
| "mean": 0.007717699743807316, |
| "std": 0.37985163927078247, |
| "abs_max": 1.40625, |
| "frac_dead": 0.00016276042151730508, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.0.op": { |
| "mean": -0.019651450216770172, |
| "std": 1.9292454719543457, |
| "abs_max": 12.5, |
| "frac_dead": 9.960598617908545e-06, |
| "frac_sat": 1.1444091796875e-05 |
| }, |
| "dec_blocks.1.norm1": { |
| "mean": -0.00014618535351473838, |
| "std": 0.9997785091400146, |
| "abs_max": 9.752230644226074, |
| "frac_dead": 9.5367431640625e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.conv1": { |
| "mean": -0.8840498328208923, |
| "std": 2.2571663856506348, |
| "abs_max": 18.625, |
| "frac_dead": 4.662407718569739e-06, |
| "frac_sat": 0.0005062951240688562 |
| }, |
| "dec_blocks.1.t_proj": { |
| "mean": -0.0025018462911248207, |
| "std": 0.04777690768241882, |
| "abs_max": 0.1845703125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.norm2": { |
| "mean": -0.002659045159816742, |
| "std": 0.997779905796051, |
| "abs_max": 8.14177417755127, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.conv2": { |
| "mean": 0.0010152194881811738, |
| "std": 1.294610619544983, |
| "abs_max": 5.3125, |
| "frac_dead": 1.1232164069951978e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.skip": { |
| "mean": 0.034126296639442444, |
| "std": 1.1006033420562744, |
| "abs_max": 7.09375, |
| "frac_dead": 3.094143539783545e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.norm": { |
| "mean": -9.974920249078423e-05, |
| "std": 0.9996920824050903, |
| "abs_max": 6.520663261413574, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.qkv": { |
| "mean": 0.013948864303529263, |
| "std": 0.7189094424247742, |
| "abs_max": 5.03125, |
| "frac_dead": 6.364892760757357e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.proj": { |
| "mean": -0.011062300764024258, |
| "std": 0.35880205035209656, |
| "abs_max": 1.65625, |
| "frac_dead": 0.00010829501843545586, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.1.op": { |
| "mean": 0.06802994757890701, |
| "std": 2.4219303131103516, |
| "abs_max": 11.8125, |
| "frac_dead": 7.841322258173022e-06, |
| "frac_sat": 1.4887916222505737e-05 |
| }, |
| "dec_blocks.2.norm1": { |
| "mean": 0.00033097798586823046, |
| "std": 1.0006484985351562, |
| "abs_max": 12.5630464553833, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 6.004615897836629e-07 |
| }, |
| "dec_blocks.2.conv1": { |
| "mean": -0.5608227252960205, |
| "std": 2.558612108230591, |
| "abs_max": 32.25, |
| "frac_dead": 8.47710543894209e-06, |
| "frac_sat": 0.006476614158600569 |
| }, |
| "dec_blocks.2.t_proj": { |
| "mean": -0.005891525652259588, |
| "std": 0.05158890783786774, |
| "abs_max": 0.2060546875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm2": { |
| "mean": -0.003271313151344657, |
| "std": 0.9989966750144958, |
| "abs_max": 12.778143882751465, |
| "frac_dead": 7.417467031700653e-07, |
| "frac_sat": 9.5367431640625e-06 |
| }, |
| "dec_blocks.2.conv2": { |
| "mean": -0.0038905921392142773, |
| "std": 0.559281587600708, |
| "abs_max": 6.3125, |
| "frac_dead": 3.804100924753584e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.skip": { |
| "mean": 0.01863963156938553, |
| "std": 1.6659446954727173, |
| "abs_max": 11.5, |
| "frac_dead": 2.5325351089122705e-05, |
| "frac_sat": 1.0596381798677612e-06 |
| }, |
| "dec_attns.2.norm": { |
| "mean": -6.945100176380947e-05, |
| "std": 0.9992446899414062, |
| "abs_max": 12.212302207946777, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 6.35782896551973e-07 |
| }, |
| "dec_attns.2.qkv": { |
| "mean": 0.013296275399625301, |
| "std": 0.6340527534484863, |
| "abs_max": 6.875, |
| "frac_dead": 9.907616913551465e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.proj": { |
| "mean": 0.01610799878835678, |
| "std": 0.27409058809280396, |
| "abs_max": 1.265625, |
| "frac_dead": 0.0002217822620877996, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.2.op": { |
| "mean": 0.1586993783712387, |
| "std": 1.4212331771850586, |
| "abs_max": 13.0, |
| "frac_dead": 1.5285280824173242e-05, |
| "frac_sat": 9.112887710216455e-06 |
| }, |
| "dec_blocks.3.norm1": { |
| "mean": 0.0002693136630114168, |
| "std": 1.0004836320877075, |
| "abs_max": 13.641996383666992, |
| "frac_dead": 6.887647714393097e-07, |
| "frac_sat": 2.370940364926355e-06 |
| }, |
| "dec_blocks.3.conv1": { |
| "mean": -0.7613422274589539, |
| "std": 1.847955346107483, |
| "abs_max": 36.0, |
| "frac_dead": 1.3536877304431982e-05, |
| "frac_sat": 0.0020664269104599953 |
| }, |
| "dec_blocks.3.t_proj": { |
| "mean": -0.0013175251660868526, |
| "std": 0.048924800008535385, |
| "abs_max": 0.1806640625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm2": { |
| "mean": -0.002513655461370945, |
| "std": 0.997054934501648, |
| "abs_max": 15.864608764648438, |
| "frac_dead": 4.76837158203125e-07, |
| "frac_sat": 2.1272235244396143e-05 |
| }, |
| "dec_blocks.3.conv2": { |
| "mean": 0.00954698771238327, |
| "std": 0.4573476016521454, |
| "abs_max": 3.921875, |
| "frac_dead": 4.7021443606354296e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.skip": { |
| "mean": 0.020214399322867393, |
| "std": 0.9816720485687256, |
| "abs_max": 11.0, |
| "frac_dead": 5.141894143889658e-05, |
| "frac_sat": 1.8543667579251633e-07 |
| }, |
| "dec_ups.3.op": { |
| "mean": 0.07220441848039627, |
| "std": 1.1501033306121826, |
| "abs_max": 13.6875, |
| "frac_dead": 2.4133258193614893e-05, |
| "frac_sat": 2.4636585749249207e-06 |
| }, |
| "dec_blocks.4.norm1": { |
| "mean": 0.0004646189045161009, |
| "std": 1.0008416175842285, |
| "abs_max": 11.96432876586914, |
| "frac_dead": 8.035589189603343e-07, |
| "frac_sat": 5.298190899338806e-07 |
| }, |
| "dec_blocks.4.conv1": { |
| "mean": -0.20519103109836578, |
| "std": 1.7457730770111084, |
| "abs_max": 21.375, |
| "frac_dead": 1.2106365829822607e-05, |
| "frac_sat": 0.0002633333206176758 |
| }, |
| "dec_blocks.4.t_proj": { |
| "mean": -0.001211022725328803, |
| "std": 0.051456209272146225, |
| "abs_max": 0.1806640625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm2": { |
| "mean": -0.001078862464055419, |
| "std": 0.9987243413925171, |
| "abs_max": 11.200509071350098, |
| "frac_dead": 1.1788474694185425e-06, |
| "frac_sat": 7.81483151968132e-07 |
| }, |
| "dec_blocks.4.conv2": { |
| "mean": -0.05559216067194939, |
| "std": 0.6134659051895142, |
| "abs_max": 9.4375, |
| "frac_dead": 7.282363367266953e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.skip": { |
| "mean": 0.0002240973262814805, |
| "std": 1.1718063354492188, |
| "abs_max": 14.0, |
| "frac_dead": 6.145900988485664e-05, |
| "frac_sat": 3.4570693969726562e-06 |
| }, |
| "dec_ups.4.op": { |
| "mean": -0.21435530483722687, |
| "std": 1.741746187210083, |
| "abs_max": 21.875, |
| "frac_dead": 2.5100178390857764e-05, |
| "frac_sat": 0.00037919150781817734 |
| }, |
| "dec_blocks.5.norm1": { |
| "mean": -0.0008769453270360827, |
| "std": 0.9995813965797424, |
| "abs_max": 13.412727355957031, |
| "frac_dead": 1.6573402490394074e-06, |
| "frac_sat": 2.880891088352655e-06 |
| }, |
| "dec_blocks.5.conv1": { |
| "mean": -0.09107916802167892, |
| "std": 1.1201961040496826, |
| "abs_max": 14.0625, |
| "frac_dead": 2.0802021026611328e-05, |
| "frac_sat": 4.119343429920264e-06 |
| }, |
| "dec_blocks.5.t_proj": { |
| "mean": -0.0012433892115950584, |
| "std": 0.05638977140188217, |
| "abs_max": 0.1982421875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.norm2": { |
| "mean": 0.0013370562810450792, |
| "std": 1.0027819871902466, |
| "abs_max": 15.821227073669434, |
| "frac_dead": 6.655852189396683e-07, |
| "frac_sat": 1.7881393432617188e-07 |
| }, |
| "dec_blocks.5.conv2": { |
| "mean": -0.7922682166099548, |
| "std": 1.2607128620147705, |
| "abs_max": 19.375, |
| "frac_dead": 3.741847103810869e-05, |
| "frac_sat": 7.07341605448164e-05 |
| }, |
| "dec_blocks.5.skip": { |
| "mean": -0.12430792301893234, |
| "std": 1.1362963914871216, |
| "abs_max": 11.875, |
| "frac_dead": 7.243951404234394e-05, |
| "frac_sat": 2.914004824106087e-07 |
| }, |
| "out_norm": { |
| "mean": -0.0004818137385882437, |
| "std": 1.0001599788665771, |
| "abs_max": 12.347586631774902, |
| "frac_dead": 1.0463926400916534e-06, |
| "frac_sat": 1.0000335350923706e-06 |
| }, |
| "out_conv": { |
| "mean": 0.4717104732990265, |
| "std": 0.23990732431411743, |
| "abs_max": 1.234375, |
| "frac_dead": 1.0596381798677612e-06, |
| "frac_sat": 0.0 |
| } |
| } |
| } |
