| { |
| "iter": 3000, |
| "per_layer": { |
| "time_mlp.1.weight": { |
| "w_norm": 19.0517578125, |
| "g_norm": 0.14865441620349884, |
| "shape": [ |
| 1024, |
| 256 |
| ] |
| }, |
| "time_mlp.1.bias": { |
| "w_norm": 1.1601998805999756, |
| "g_norm": 0.013137980364263058, |
| "shape": [ |
| 1024 |
| ] |
| }, |
| "time_mlp.3.weight": { |
| "w_norm": 9.77436351776123, |
| "g_norm": 0.022276513278484344, |
| "shape": [ |
| 256, |
| 1024 |
| ] |
| }, |
| "time_mlp.3.bias": { |
| "w_norm": 0.2890886664390564, |
| "g_norm": 0.011864181607961655, |
| "shape": [ |
| 256 |
| ] |
| }, |
| "input_conv.weight": { |
| "w_norm": 7.956737995147705, |
| "g_norm": 0.19501574337482452, |
| "shape": [ |
| 192, |
| 6, |
| 3, |
| 3 |
| ] |
| }, |
| "input_conv.bias": { |
| "w_norm": 1.1898951530456543, |
| "g_norm": 0.19526450335979462, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.weight": { |
| "w_norm": 13.789409637451172, |
| "g_norm": 0.05440613627433777, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm1.bias": { |
| "w_norm": 0.12104981392621994, |
| "g_norm": 0.08626538515090942, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv1.weight": { |
| "w_norm": 8.61393928527832, |
| "g_norm": 0.4161059260368347, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv1.bias": { |
| "w_norm": 0.20222757756710052, |
| "g_norm": 0.04245798662304878, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.t_proj.weight": { |
| "w_norm": 8.108176231384277, |
| "g_norm": 0.04685848206281662, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.0.t_proj.bias": { |
| "w_norm": 0.5027391910552979, |
| "g_norm": 0.042459018528461456, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.weight": { |
| "w_norm": 13.823328971862793, |
| "g_norm": 0.026325175538659096, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.norm2.bias": { |
| "w_norm": 0.08441212773323059, |
| "g_norm": 0.033365774899721146, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.0.conv2.weight": { |
| "w_norm": 8.440913200378418, |
| "g_norm": 0.1690007895231247, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.0.conv2.bias": { |
| "w_norm": 0.18897441029548645, |
| "g_norm": 0.01589149981737137, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.weight": { |
| "w_norm": 13.84272575378418, |
| "g_norm": 0.0005544356536120176, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm1.bias": { |
| "w_norm": 0.053813960403203964, |
| "g_norm": 0.000708088162355125, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv1.weight": { |
| "w_norm": 8.352360725402832, |
| "g_norm": 0.009228352457284927, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv1.bias": { |
| "w_norm": 0.1925705522298813, |
| "g_norm": 0.0006506195059046149, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.t_proj.weight": { |
| "w_norm": 8.264411926269531, |
| "g_norm": 0.0007487694965675473, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "enc_blocks.1.t_proj.bias": { |
| "w_norm": 0.523145854473114, |
| "g_norm": 0.0006508316146209836, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.weight": { |
| "w_norm": 13.815474510192871, |
| "g_norm": 0.0007219835533760488, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.norm2.bias": { |
| "w_norm": 0.08295512944459915, |
| "g_norm": 0.0007739425054751337, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.1.conv2.weight": { |
| "w_norm": 8.333471298217773, |
| "g_norm": 0.012200208380818367, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.1.conv2.bias": { |
| "w_norm": 0.20990578830242157, |
| "g_norm": 0.0010111451847478747, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.weight": { |
| "w_norm": 13.804555892944336, |
| "g_norm": 0.0002658558951225132, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.norm1.bias": { |
| "w_norm": 0.07251661270856857, |
| "g_norm": 0.00028446578653529286, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_blocks.2.conv1.weight": { |
| "w_norm": 11.796462059020996, |
| "g_norm": 0.00543182622641325, |
| "shape": [ |
| 384, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv1.bias": { |
| "w_norm": 0.26768797636032104, |
| "g_norm": 0.00036368807195685804, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.t_proj.weight": { |
| "w_norm": 12.015633583068848, |
| "g_norm": 0.0003876331902574748, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.2.t_proj.bias": { |
| "w_norm": 0.7215366363525391, |
| "g_norm": 0.00036360954982228577, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.weight": { |
| "w_norm": 19.397293090820312, |
| "g_norm": 0.0003247189160902053, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.norm2.bias": { |
| "w_norm": 0.26001405715942383, |
| "g_norm": 0.00037615434848703444, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.conv2.weight": { |
| "w_norm": 11.833823204040527, |
| "g_norm": 0.009932780638337135, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.2.conv2.bias": { |
| "w_norm": 0.20211109519004822, |
| "g_norm": 0.0006463375175371766, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.2.skip.weight": { |
| "w_norm": 11.166708946228027, |
| "g_norm": 0.010496451519429684, |
| "shape": [ |
| 384, |
| 192, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.2.skip.bias": { |
| "w_norm": 0.8282492756843567, |
| "g_norm": 0.0006463375175371766, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.weight": { |
| "w_norm": 19.57082176208496, |
| "g_norm": 0.00017050265159923583, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm1.bias": { |
| "w_norm": 0.07634127885103226, |
| "g_norm": 0.000175074819708243, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv1.weight": { |
| "w_norm": 12.135543823242188, |
| "g_norm": 0.003038321156054735, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv1.bias": { |
| "w_norm": 0.19764570891857147, |
| "g_norm": 0.00012538516602944583, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.t_proj.weight": { |
| "w_norm": 11.44118881225586, |
| "g_norm": 0.00013460988702718168, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "enc_blocks.3.t_proj.bias": { |
| "w_norm": 0.692889392375946, |
| "g_norm": 0.00012540096940938383, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.weight": { |
| "w_norm": 19.37941551208496, |
| "g_norm": 0.0001833721762523055, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.norm2.bias": { |
| "w_norm": 0.2353246659040451, |
| "g_norm": 0.00028943736106157303, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.3.conv2.weight": { |
| "w_norm": 11.664400100708008, |
| "g_norm": 0.006047896109521389, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.3.conv2.bias": { |
| "w_norm": 0.1913631409406662, |
| "g_norm": 0.0004047580005135387, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.weight": { |
| "w_norm": 19.603595733642578, |
| "g_norm": 0.0004483657539822161, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.norm1.bias": { |
| "w_norm": 0.07635044306516647, |
| "g_norm": 0.00047479913337156177, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_blocks.4.conv1.weight": { |
| "w_norm": 16.962387084960938, |
| "g_norm": 0.010729203931987286, |
| "shape": [ |
| 768, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv1.bias": { |
| "w_norm": 0.2939714789390564, |
| "g_norm": 0.00041833348223008215, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.t_proj.weight": { |
| "w_norm": 16.067256927490234, |
| "g_norm": 0.00043840549187734723, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.4.t_proj.bias": { |
| "w_norm": 0.9943748712539673, |
| "g_norm": 0.00041832224815152586, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.weight": { |
| "w_norm": 27.357318878173828, |
| "g_norm": 0.0007662373827770352, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.norm2.bias": { |
| "w_norm": 0.4109998047351837, |
| "g_norm": 0.0009103051270358264, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.conv2.weight": { |
| "w_norm": 16.50996208190918, |
| "g_norm": 0.011206631548702717, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.4.conv2.bias": { |
| "w_norm": 0.19958554208278656, |
| "g_norm": 0.00041942455573007464, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.4.skip.weight": { |
| "w_norm": 15.74730110168457, |
| "g_norm": 0.011705640703439713, |
| "shape": [ |
| 768, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_blocks.4.skip.bias": { |
| "w_norm": 0.8086597919464111, |
| "g_norm": 0.00041942455573007464, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.weight": { |
| "w_norm": 27.715965270996094, |
| "g_norm": 0.00013690636842511594, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm1.bias": { |
| "w_norm": 0.07511666417121887, |
| "g_norm": 0.00011273950076429173, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv1.weight": { |
| "w_norm": 17.259342193603516, |
| "g_norm": 0.002783330390229821, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv1.bias": { |
| "w_norm": 0.20967231690883636, |
| "g_norm": 5.966354729025625e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.t_proj.weight": { |
| "w_norm": 16.016868591308594, |
| "g_norm": 6.135379226179793e-05, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "enc_blocks.5.t_proj.bias": { |
| "w_norm": 1.0008527040481567, |
| "g_norm": 5.964525553281419e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.weight": { |
| "w_norm": 27.47331428527832, |
| "g_norm": 0.00027751424931921065, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.norm2.bias": { |
| "w_norm": 0.28648436069488525, |
| "g_norm": 0.00025248766178265214, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_blocks.5.conv2.weight": { |
| "w_norm": 16.705930709838867, |
| "g_norm": 0.007712048012763262, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_blocks.5.conv2.bias": { |
| "w_norm": 0.19320331513881683, |
| "g_norm": 0.0002017767692450434, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.3.norm.weight": { |
| "w_norm": 19.50695037841797, |
| "g_norm": 9.197030158247799e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.norm.bias": { |
| "w_norm": 0.04814915731549263, |
| "g_norm": 0.00017879036022350192, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.3.qkv.weight": { |
| "w_norm": 19.677698135375977, |
| "g_norm": 0.0026700582820922136, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.qkv.bias": { |
| "w_norm": 1.0045832395553589, |
| "g_norm": 0.0002563674352131784, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "enc_attns.3.proj.weight": { |
| "w_norm": 11.314918518066406, |
| "g_norm": 0.002640425693243742, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.3.proj.bias": { |
| "w_norm": 0.561999499797821, |
| "g_norm": 0.0004093097522854805, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_attns.4.norm.weight": { |
| "w_norm": 27.55660629272461, |
| "g_norm": 0.00014296467998065054, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.norm.bias": { |
| "w_norm": 0.07683202624320984, |
| "g_norm": 0.00021652992290910333, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.4.qkv.weight": { |
| "w_norm": 28.08203887939453, |
| "g_norm": 0.004884236492216587, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.qkv.bias": { |
| "w_norm": 1.0157814025878906, |
| "g_norm": 0.0002601943851914257, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.4.proj.weight": { |
| "w_norm": 16.036779403686523, |
| "g_norm": 0.004596893209964037, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.4.proj.bias": { |
| "w_norm": 0.5738911628723145, |
| "g_norm": 0.000422380689997226, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.weight": { |
| "w_norm": 27.647903442382812, |
| "g_norm": 4.3613141315290704e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.norm.bias": { |
| "w_norm": 0.06394332647323608, |
| "g_norm": 9.029910142999142e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_attns.5.qkv.weight": { |
| "w_norm": 28.060827255249023, |
| "g_norm": 0.0017676395364105701, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.qkv.bias": { |
| "w_norm": 1.0001214742660522, |
| "g_norm": 0.00012707892165053636, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "enc_attns.5.proj.weight": { |
| "w_norm": 16.089628219604492, |
| "g_norm": 0.001794372103177011, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "enc_attns.5.proj.bias": { |
| "w_norm": 0.5971724987030029, |
| "g_norm": 0.0002005139976972714, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "enc_downs.0.op.weight": { |
| "w_norm": 8.235817909240723, |
| "g_norm": 0.018652230501174927, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.0.op.bias": { |
| "w_norm": 0.18452279269695282, |
| "g_norm": 0.0009377316455356777, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.1.op.weight": { |
| "w_norm": 8.140389442443848, |
| "g_norm": 0.017915265634655952, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.1.op.bias": { |
| "w_norm": 0.19395767152309418, |
| "g_norm": 0.0005633846158161759, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "enc_downs.2.op.weight": { |
| "w_norm": 11.588905334472656, |
| "g_norm": 0.016342995688319206, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.2.op.bias": { |
| "w_norm": 0.18809567391872406, |
| "g_norm": 0.00040573853766545653, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.3.op.weight": { |
| "w_norm": 11.490419387817383, |
| "g_norm": 0.022241534665226936, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.3.op.bias": { |
| "w_norm": 0.19505050778388977, |
| "g_norm": 0.00036343102692626417, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "enc_downs.4.op.weight": { |
| "w_norm": 16.517168045043945, |
| "g_norm": 0.023726537823677063, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "enc_downs.4.op.bias": { |
| "w_norm": 0.20217803120613098, |
| "g_norm": 0.00020157717517577112, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.weight": { |
| "w_norm": 27.712066650390625, |
| "g_norm": 7.81923154136166e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm1.bias": { |
| "w_norm": 0.05954904481768608, |
| "g_norm": 6.75633127684705e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv1.weight": { |
| "w_norm": 17.084714889526367, |
| "g_norm": 0.0018813041970133781, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv1.bias": { |
| "w_norm": 0.20746949315071106, |
| "g_norm": 4.295578764867969e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.t_proj.weight": { |
| "w_norm": 16.020248413085938, |
| "g_norm": 4.511191582423635e-05, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid1.t_proj.bias": { |
| "w_norm": 1.030643105506897, |
| "g_norm": 4.29449173680041e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.weight": { |
| "w_norm": 27.478395462036133, |
| "g_norm": 5.8335594076197594e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.norm2.bias": { |
| "w_norm": 0.32949304580688477, |
| "g_norm": 8.398802310694009e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid1.conv2.weight": { |
| "w_norm": 16.644107818603516, |
| "g_norm": 0.002602463588118553, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid1.conv2.bias": { |
| "w_norm": 0.19403482973575592, |
| "g_norm": 0.00012408455950208008, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.weight": { |
| "w_norm": 27.65032958984375, |
| "g_norm": 2.737711838562973e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.norm.bias": { |
| "w_norm": 0.0695677399635315, |
| "g_norm": 4.668824476539157e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid_attn.qkv.weight": { |
| "w_norm": 27.996524810791016, |
| "g_norm": 0.001264288672246039, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.qkv.bias": { |
| "w_norm": 0.9832267761230469, |
| "g_norm": 7.399465539492667e-05, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "mid_attn.proj.weight": { |
| "w_norm": 16.07176399230957, |
| "g_norm": 0.0011793198063969612, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "mid_attn.proj.bias": { |
| "w_norm": 0.576002836227417, |
| "g_norm": 0.00012308346049394459, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.weight": { |
| "w_norm": 27.723308563232422, |
| "g_norm": 0.00023281383619178087, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm1.bias": { |
| "w_norm": 0.06764264404773712, |
| "g_norm": 0.00020761771884281188, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv1.weight": { |
| "w_norm": 17.205245971679688, |
| "g_norm": 0.0032772363629192114, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv1.bias": { |
| "w_norm": 0.2025199979543686, |
| "g_norm": 7.581734098494053e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.t_proj.weight": { |
| "w_norm": 16.00699234008789, |
| "g_norm": 7.93769650044851e-05, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "mid2.t_proj.bias": { |
| "w_norm": 0.9997656941413879, |
| "g_norm": 7.582550460938364e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.weight": { |
| "w_norm": 27.45841407775879, |
| "g_norm": 0.0001569108135299757, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.norm2.bias": { |
| "w_norm": 0.3297288417816162, |
| "g_norm": 0.00017398105410393327, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "mid2.conv2.weight": { |
| "w_norm": 16.620267868041992, |
| "g_norm": 0.0026973795611411333, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "mid2.conv2.bias": { |
| "w_norm": 0.19106969237327576, |
| "g_norm": 0.00011876138887600973, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm1.weight": { |
| "w_norm": 39.22458267211914, |
| "g_norm": 0.0001079371286323294, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.norm1.bias": { |
| "w_norm": 0.1136740893125534, |
| "g_norm": 9.537713776808232e-05, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.0.conv1.weight": { |
| "w_norm": 18.268850326538086, |
| "g_norm": 0.0026090722531080246, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv1.bias": { |
| "w_norm": 0.15742380917072296, |
| "g_norm": 4.251589416526258e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.t_proj.weight": { |
| "w_norm": 16.02547836303711, |
| "g_norm": 4.457262184587307e-05, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.0.t_proj.bias": { |
| "w_norm": 1.0245296955108643, |
| "g_norm": 4.250950587447733e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.weight": { |
| "w_norm": 27.40232276916504, |
| "g_norm": 0.00011834588804049417, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.norm2.bias": { |
| "w_norm": 0.4019148647785187, |
| "g_norm": 0.00021984959312248975, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.conv2.weight": { |
| "w_norm": 16.37566566467285, |
| "g_norm": 0.0035729294177144766, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.0.conv2.bias": { |
| "w_norm": 0.19908437132835388, |
| "g_norm": 0.00025187665596604347, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.0.skip.weight": { |
| "w_norm": 15.740446090698242, |
| "g_norm": 0.037088118493556976, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.0.skip.bias": { |
| "w_norm": 0.4123978316783905, |
| "g_norm": 0.00025187665596604347, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm1.weight": { |
| "w_norm": 39.235748291015625, |
| "g_norm": 0.00011050483590224758, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.norm1.bias": { |
| "w_norm": 0.10630356520414352, |
| "g_norm": 0.00011029136658180505, |
| "shape": [ |
| 1536 |
| ] |
| }, |
| "dec_blocks.1.conv1.weight": { |
| "w_norm": 18.143503189086914, |
| "g_norm": 0.0031994630116969347, |
| "shape": [ |
| 768, |
| 1536, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv1.bias": { |
| "w_norm": 0.1565992832183838, |
| "g_norm": 5.2350569603731856e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.t_proj.weight": { |
| "w_norm": 16.061790466308594, |
| "g_norm": 5.610016523860395e-05, |
| "shape": [ |
| 768, |
| 256 |
| ] |
| }, |
| "dec_blocks.1.t_proj.bias": { |
| "w_norm": 0.9904308915138245, |
| "g_norm": 5.23491544299759e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.weight": { |
| "w_norm": 27.39215660095215, |
| "g_norm": 9.991656406782568e-05, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.norm2.bias": { |
| "w_norm": 0.43167370557785034, |
| "g_norm": 0.0002531306236051023, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.conv2.weight": { |
| "w_norm": 16.279890060424805, |
| "g_norm": 0.003765431232750416, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.1.conv2.bias": { |
| "w_norm": 0.19697929918766022, |
| "g_norm": 0.0002718368486966938, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.1.skip.weight": { |
| "w_norm": 15.701605796813965, |
| "g_norm": 0.051730845123529434, |
| "shape": [ |
| 768, |
| 1536, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.1.skip.bias": { |
| "w_norm": 0.407044380903244, |
| "g_norm": 0.0002718368486966938, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.2.norm1.weight": { |
| "w_norm": 33.96385192871094, |
| "g_norm": 3.643078525783494e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.norm1.bias": { |
| "w_norm": 0.10771745443344116, |
| "g_norm": 3.162841312587261e-05, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_blocks.2.conv1.weight": { |
| "w_norm": 12.829668045043945, |
| "g_norm": 0.0013110062573105097, |
| "shape": [ |
| 384, |
| 1152, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv1.bias": { |
| "w_norm": 0.12390010058879852, |
| "g_norm": 2.300715095771011e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.t_proj.weight": { |
| "w_norm": 11.354249954223633, |
| "g_norm": 2.4828577807056718e-05, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.2.t_proj.bias": { |
| "w_norm": 0.7089831829071045, |
| "g_norm": 2.300190317328088e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.weight": { |
| "w_norm": 19.416515350341797, |
| "g_norm": 5.5451666412409395e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.norm2.bias": { |
| "w_norm": 0.26346495747566223, |
| "g_norm": 9.812486678129062e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.conv2.weight": { |
| "w_norm": 11.493674278259277, |
| "g_norm": 0.003377738408744335, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.2.conv2.bias": { |
| "w_norm": 0.18863797187805176, |
| "g_norm": 0.0002339720376767218, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.2.skip.weight": { |
| "w_norm": 10.904874801635742, |
| "g_norm": 0.04481441527605057, |
| "shape": [ |
| 384, |
| 1152, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.2.skip.bias": { |
| "w_norm": 0.3309810757637024, |
| "g_norm": 0.0002339720376767218, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm1.weight": { |
| "w_norm": 27.707651138305664, |
| "g_norm": 0.00013387756189331412, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.norm1.bias": { |
| "w_norm": 0.09725356847047806, |
| "g_norm": 0.00011039790115319192, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_blocks.3.conv1.weight": { |
| "w_norm": 12.639841079711914, |
| "g_norm": 0.004064203705638647, |
| "shape": [ |
| 384, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv1.bias": { |
| "w_norm": 0.14569738507270813, |
| "g_norm": 9.590496483724564e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.t_proj.weight": { |
| "w_norm": 11.424116134643555, |
| "g_norm": 0.00010441212361911312, |
| "shape": [ |
| 384, |
| 256 |
| ] |
| }, |
| "dec_blocks.3.t_proj.bias": { |
| "w_norm": 0.7090743184089661, |
| "g_norm": 9.588411921868101e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.weight": { |
| "w_norm": 19.390981674194336, |
| "g_norm": 0.00021730814478360116, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.norm2.bias": { |
| "w_norm": 0.2664073705673218, |
| "g_norm": 0.00026408053236082196, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.conv2.weight": { |
| "w_norm": 11.58509349822998, |
| "g_norm": 0.008606428280472755, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.3.conv2.bias": { |
| "w_norm": 0.19408394396305084, |
| "g_norm": 0.0005706407828256488, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.3.skip.weight": { |
| "w_norm": 11.021206855773926, |
| "g_norm": 0.05618388578295708, |
| "shape": [ |
| 384, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.3.skip.bias": { |
| "w_norm": 0.4036846458911896, |
| "g_norm": 0.0005706407828256488, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.4.norm1.weight": { |
| "w_norm": 23.96885108947754, |
| "g_norm": 0.00025959074264392257, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.norm1.bias": { |
| "w_norm": 0.11969521641731262, |
| "g_norm": 0.00022613562759943306, |
| "shape": [ |
| 576 |
| ] |
| }, |
| "dec_blocks.4.conv1.weight": { |
| "w_norm": 8.764872550964355, |
| "g_norm": 0.009391983039677143, |
| "shape": [ |
| 192, |
| 576, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv1.bias": { |
| "w_norm": 0.11739768087863922, |
| "g_norm": 0.00023739114112686366, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.t_proj.weight": { |
| "w_norm": 8.095752716064453, |
| "g_norm": 0.00026289356173947453, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.4.t_proj.bias": { |
| "w_norm": 0.4813956618309021, |
| "g_norm": 0.0002374465111643076, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.weight": { |
| "w_norm": 13.783257484436035, |
| "g_norm": 0.0007113862666301429, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.norm2.bias": { |
| "w_norm": 0.1056097000837326, |
| "g_norm": 0.0006580191547982395, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.conv2.weight": { |
| "w_norm": 8.158761978149414, |
| "g_norm": 0.02111119031906128, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.4.conv2.bias": { |
| "w_norm": 0.19651232659816742, |
| "g_norm": 0.0013463678769767284, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.4.skip.weight": { |
| "w_norm": 7.813741683959961, |
| "g_norm": 0.09076514840126038, |
| "shape": [ |
| 192, |
| 576, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.4.skip.bias": { |
| "w_norm": 0.3467942774295807, |
| "g_norm": 0.0013463678769767284, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm1.weight": { |
| "w_norm": 19.519624710083008, |
| "g_norm": 0.011791425757110119, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.norm1.bias": { |
| "w_norm": 0.14571304619312286, |
| "g_norm": 0.013299982994794846, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_blocks.5.conv1.weight": { |
| "w_norm": 8.778403282165527, |
| "g_norm": 0.08446153998374939, |
| "shape": [ |
| 192, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv1.bias": { |
| "w_norm": 0.13901221752166748, |
| "g_norm": 0.00463363341987133, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.t_proj.weight": { |
| "w_norm": 8.244766235351562, |
| "g_norm": 0.005314541980624199, |
| "shape": [ |
| 192, |
| 256 |
| ] |
| }, |
| "dec_blocks.5.t_proj.bias": { |
| "w_norm": 0.5051465034484863, |
| "g_norm": 0.0046348548494279385, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.weight": { |
| "w_norm": 13.902420043945312, |
| "g_norm": 0.0054893940687179565, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.norm2.bias": { |
| "w_norm": 0.04917939752340317, |
| "g_norm": 0.0060037486255168915, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.conv2.weight": { |
| "w_norm": 8.520387649536133, |
| "g_norm": 0.0472685769200325, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_blocks.5.conv2.bias": { |
| "w_norm": 0.2030690759420395, |
| "g_norm": 0.0030816341750323772, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_blocks.5.skip.weight": { |
| "w_norm": 8.117816925048828, |
| "g_norm": 0.08592767268419266, |
| "shape": [ |
| 192, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_blocks.5.skip.bias": { |
| "w_norm": 0.42368704080581665, |
| "g_norm": 0.0030816341750323772, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "dec_attns.0.norm.weight": { |
| "w_norm": 27.58348846435547, |
| "g_norm": 0.00026784223155118525, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.norm.bias": { |
| "w_norm": 0.05455893278121948, |
| "g_norm": 0.00027358514489606023, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.0.qkv.weight": { |
| "w_norm": 28.11470603942871, |
| "g_norm": 0.006553055718541145, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.qkv.bias": { |
| "w_norm": 1.0048754215240479, |
| "g_norm": 0.00023312140547204763, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.0.proj.weight": { |
| "w_norm": 16.170562744140625, |
| "g_norm": 0.005232008174061775, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.0.proj.bias": { |
| "w_norm": 0.5775165557861328, |
| "g_norm": 0.00028337317053228617, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.weight": { |
| "w_norm": 27.570045471191406, |
| "g_norm": 0.000315696292091161, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.norm.bias": { |
| "w_norm": 0.05546895042061806, |
| "g_norm": 0.0002729733823798597, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.1.qkv.weight": { |
| "w_norm": 28.15424156188965, |
| "g_norm": 0.008273596875369549, |
| "shape": [ |
| 2304, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.qkv.bias": { |
| "w_norm": 1.0058194398880005, |
| "g_norm": 0.00025404043844901025, |
| "shape": [ |
| 2304 |
| ] |
| }, |
| "dec_attns.1.proj.weight": { |
| "w_norm": 16.085893630981445, |
| "g_norm": 0.007718106731772423, |
| "shape": [ |
| 768, |
| 768, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.1.proj.bias": { |
| "w_norm": 0.5793389081954956, |
| "g_norm": 0.0003180239873472601, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_attns.2.norm.weight": { |
| "w_norm": 19.477664947509766, |
| "g_norm": 9.047827188624069e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.norm.bias": { |
| "w_norm": 0.031263552606105804, |
| "g_norm": 8.09360426501371e-05, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_attns.2.qkv.weight": { |
| "w_norm": 19.72080421447754, |
| "g_norm": 0.002651409013196826, |
| "shape": [ |
| 1152, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.qkv.bias": { |
| "w_norm": 0.9826018810272217, |
| "g_norm": 0.00012897119449917227, |
| "shape": [ |
| 1152 |
| ] |
| }, |
| "dec_attns.2.proj.weight": { |
| "w_norm": 11.312171936035156, |
| "g_norm": 0.002945007523521781, |
| "shape": [ |
| 384, |
| 384, |
| 1, |
| 1 |
| ] |
| }, |
| "dec_attns.2.proj.bias": { |
| "w_norm": 0.5890069007873535, |
| "g_norm": 0.00023511593462899327, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.0.op.weight": { |
| "w_norm": 16.251949310302734, |
| "g_norm": 0.02308141626417637, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.0.op.bias": { |
| "w_norm": 0.20316992700099945, |
| "g_norm": 0.00010308789205737412, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.1.op.weight": { |
| "w_norm": 16.251108169555664, |
| "g_norm": 0.03540193662047386, |
| "shape": [ |
| 768, |
| 768, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.1.op.bias": { |
| "w_norm": 0.1951359361410141, |
| "g_norm": 0.0001347730722045526, |
| "shape": [ |
| 768 |
| ] |
| }, |
| "dec_ups.2.op.weight": { |
| "w_norm": 11.344901084899902, |
| "g_norm": 0.06848753243684769, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.2.op.bias": { |
| "w_norm": 0.19678521156311035, |
| "g_norm": 0.00026137454551644623, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.3.op.weight": { |
| "w_norm": 11.392304420471191, |
| "g_norm": 0.0678146630525589, |
| "shape": [ |
| 384, |
| 384, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.3.op.bias": { |
| "w_norm": 0.2005225270986557, |
| "g_norm": 0.0004707111802417785, |
| "shape": [ |
| 384 |
| ] |
| }, |
| "dec_ups.4.op.weight": { |
| "w_norm": 8.021990776062012, |
| "g_norm": 0.0939137414097786, |
| "shape": [ |
| 192, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "dec_ups.4.op.bias": { |
| "w_norm": 0.19830958545207977, |
| "g_norm": 0.0010575436754152179, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.weight": { |
| "w_norm": 13.740449905395508, |
| "g_norm": 0.009818742983043194, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_norm.bias": { |
| "w_norm": 0.13754217326641083, |
| "g_norm": 0.013854917138814926, |
| "shape": [ |
| 192 |
| ] |
| }, |
| "out_conv.weight": { |
| "w_norm": 0.15520590543746948, |
| "g_norm": 0.785935640335083, |
| "shape": [ |
| 3, |
| 192, |
| 3, |
| 3 |
| ] |
| }, |
| "out_conv.bias": { |
| "w_norm": 0.009202906861901283, |
| "g_norm": 0.06661342829465866, |
| "shape": [ |
| 3 |
| ] |
| } |
| }, |
| "activations": { |
| "input_conv": { |
| "mean": 0.007421940565109253, |
| "std": 0.45184287428855896, |
| "abs_max": 3.21875, |
| "frac_dead": 0.00039866898441687226, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.1": { |
| "mean": 0.007643388584256172, |
| "std": 0.2923099100589752, |
| "abs_max": 4.53125, |
| "frac_dead": 4.069010537932627e-05, |
| "frac_sat": 0.0 |
| }, |
| "time_mlp.3": { |
| "mean": 0.025249801576137543, |
| "std": 0.23188936710357666, |
| "abs_max": 1.1484375, |
| "frac_dead": 0.00016276042151730508, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm1": { |
| "mean": -0.007203842978924513, |
| "std": 0.9947881102561951, |
| "abs_max": 6.314060688018799, |
| "frac_dead": 4.072984154390724e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv1": { |
| "mean": 0.1297854334115982, |
| "std": 1.0878294706344604, |
| "abs_max": 10.1875, |
| "frac_dead": 3.8295984268188477e-05, |
| "frac_sat": 3.311369178859991e-09 |
| }, |
| "enc_blocks.0.t_proj": { |
| "mean": 0.0019384645856916904, |
| "std": 0.2523488402366638, |
| "abs_max": 2.78125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.norm2": { |
| "mean": -0.005775494500994682, |
| "std": 0.9985000491142273, |
| "abs_max": 9.170462608337402, |
| "frac_dead": 8.245309004450974e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.0.conv2": { |
| "mean": -0.035144317895174026, |
| "std": 1.0234105587005615, |
| "abs_max": 13.1875, |
| "frac_dead": 3.355410444783047e-05, |
| "frac_sat": 1.430511474609375e-06 |
| }, |
| "enc_downs.0.op": { |
| "mean": -0.014735549688339233, |
| "std": 2.3309483528137207, |
| "abs_max": 33.25, |
| "frac_dead": 1.627869096409995e-05, |
| "frac_sat": 0.0026156241074204445 |
| }, |
| "enc_blocks.1.norm1": { |
| "mean": -0.0033612640108913183, |
| "std": 0.999249279499054, |
| "abs_max": 13.021814346313477, |
| "frac_dead": 5.828009648212173e-07, |
| "frac_sat": 1.5894572413799324e-07 |
| }, |
| "enc_blocks.1.conv1": { |
| "mean": 0.024152027443051338, |
| "std": 1.0423327684402466, |
| "abs_max": 15.3125, |
| "frac_dead": 3.337860107421875e-05, |
| "frac_sat": 7.973777428560425e-06 |
| }, |
| "enc_blocks.1.t_proj": { |
| "mean": 0.006402192637324333, |
| "std": 0.3085084557533264, |
| "abs_max": 2.359375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.1.norm2": { |
| "mean": -0.00576779106631875, |
| "std": 0.9974737763404846, |
| "abs_max": 15.784193992614746, |
| "frac_dead": 2.3444492853741394e-06, |
| "frac_sat": 1.4702478665640228e-06 |
| }, |
| "enc_blocks.1.conv2": { |
| "mean": -0.022583598271012306, |
| "std": 0.932984471321106, |
| "abs_max": 12.3125, |
| "frac_dead": 3.710058081196621e-05, |
| "frac_sat": 3.8411883451772155e-07 |
| }, |
| "enc_downs.1.op": { |
| "mean": -0.06834577023983002, |
| "std": 2.8865416049957275, |
| "abs_max": 37.0, |
| "frac_dead": 1.2132856682001147e-05, |
| "frac_sat": 0.006111780647188425 |
| }, |
| "enc_blocks.2.norm1": { |
| "mean": -0.004561815410852432, |
| "std": 0.9969245791435242, |
| "abs_max": 9.989190101623535, |
| "frac_dead": 2.649095449669403e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.conv1": { |
| "mean": -0.1135186105966568, |
| "std": 1.2132693529129028, |
| "abs_max": 18.625, |
| "frac_dead": 4.283587259124033e-05, |
| "frac_sat": 0.00017280048632528633 |
| }, |
| "enc_blocks.2.t_proj": { |
| "mean": -0.0073701548390090466, |
| "std": 0.4103415012359619, |
| "abs_max": 3.734375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.2.norm2": { |
| "mean": -0.014336032792925835, |
| "std": 0.9936417937278748, |
| "abs_max": 12.799908638000488, |
| "frac_dead": 5.033281240685028e-07, |
| "frac_sat": 7.523430667788489e-06 |
| }, |
| "enc_blocks.2.conv2": { |
| "mean": -0.015473869629204273, |
| "std": 1.0459991693496704, |
| "abs_max": 14.1875, |
| "frac_dead": 2.0371542632346973e-05, |
| "frac_sat": 2.9934776648588013e-06 |
| }, |
| "enc_blocks.2.skip": { |
| "mean": -0.03707383945584297, |
| "std": 1.4708917140960693, |
| "abs_max": 17.25, |
| "frac_dead": 7.65588556532748e-05, |
| "frac_sat": 4.4345855712890625e-05 |
| }, |
| "enc_downs.2.op": { |
| "mean": -0.27560582756996155, |
| "std": 2.0658843517303467, |
| "abs_max": 35.0, |
| "frac_dead": 1.0914272934314795e-05, |
| "frac_sat": 0.0011461046524345875 |
| }, |
| "enc_blocks.3.norm1": { |
| "mean": -0.002862681169062853, |
| "std": 0.9991037249565125, |
| "abs_max": 9.527902603149414, |
| "frac_dead": 6.35782896551973e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.conv1": { |
| "mean": -0.22325478494167328, |
| "std": 1.7933238744735718, |
| "abs_max": 34.75, |
| "frac_dead": 1.706017428659834e-05, |
| "frac_sat": 0.0021811590995639563 |
| }, |
| "enc_blocks.3.t_proj": { |
| "mean": 0.001510739792138338, |
| "std": 0.1843785047531128, |
| "abs_max": 1.34375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.3.norm2": { |
| "mean": -0.01221651490777731, |
| "std": 0.990524411201477, |
| "abs_max": 19.570201873779297, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 4.76837158203125e-05 |
| }, |
| "enc_blocks.3.conv2": { |
| "mean": 0.11187144368886948, |
| "std": 1.2421133518218994, |
| "abs_max": 11.3125, |
| "frac_dead": 2.3312039047596045e-05, |
| "frac_sat": 1.9073486328125e-06 |
| }, |
| "enc_attns.3.norm": { |
| "mean": 0.00036654993891716003, |
| "std": 0.9955237507820129, |
| "abs_max": 10.021247863769531, |
| "frac_dead": 9.5367431640625e-07, |
| "frac_sat": 1.059638137235197e-07 |
| }, |
| "enc_attns.3.qkv": { |
| "mean": 0.005369497928768396, |
| "std": 0.6125555634498596, |
| "abs_max": 6.71875, |
| "frac_dead": 0.00010663492139428854, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.3.proj": { |
| "mean": 0.0026778418105095625, |
| "std": 0.3171274662017822, |
| "abs_max": 1.953125, |
| "frac_dead": 0.00018797980737872422, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.3.op": { |
| "mean": -0.23350858688354492, |
| "std": 2.7067246437072754, |
| "abs_max": 41.25, |
| "frac_dead": 1.229180270456709e-05, |
| "frac_sat": 0.004200829658657312 |
| }, |
| "enc_blocks.4.norm1": { |
| "mean": -0.00010409951210021973, |
| "std": 1.0004417896270752, |
| "abs_max": 9.303688049316406, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.conv1": { |
| "mean": -0.30328503251075745, |
| "std": 1.6799970865249634, |
| "abs_max": 31.625, |
| "frac_dead": 2.373589450144209e-05, |
| "frac_sat": 0.0029724969062954187 |
| }, |
| "enc_blocks.4.t_proj": { |
| "mean": -0.0012046759948134422, |
| "std": 0.11991295218467712, |
| "abs_max": 1.203125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.norm2": { |
| "mean": -0.01589149236679077, |
| "std": 0.9891142845153809, |
| "abs_max": 17.085819244384766, |
| "frac_dead": 1.0596381798677612e-06, |
| "frac_sat": 6.76049166941084e-05 |
| }, |
| "enc_blocks.4.conv2": { |
| "mean": 0.14119984209537506, |
| "std": 1.3356584310531616, |
| "abs_max": 8.9375, |
| "frac_dead": 1.229180270456709e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.4.skip": { |
| "mean": -0.0065156240016222, |
| "std": 1.5490919351577759, |
| "abs_max": 31.0, |
| "frac_dead": 4.789564263774082e-05, |
| "frac_sat": 0.00041135153151117265 |
| }, |
| "enc_attns.4.norm": { |
| "mean": 0.00013783077883999795, |
| "std": 0.994690477848053, |
| "abs_max": 10.197798728942871, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 4.238552548940788e-07 |
| }, |
| "enc_attns.4.qkv": { |
| "mean": 0.012843161821365356, |
| "std": 0.6926348209381104, |
| "abs_max": 8.1875, |
| "frac_dead": 6.852326623629779e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.4.proj": { |
| "mean": 0.002519653644412756, |
| "std": 0.4470694065093994, |
| "abs_max": 5.0, |
| "frac_dead": 0.00011083814752055332, |
| "frac_sat": 0.0 |
| }, |
| "enc_downs.4.op": { |
| "mean": 0.2122226357460022, |
| "std": 3.1655666828155518, |
| "abs_max": 45.0, |
| "frac_dead": 3.3908420391526306e-06, |
| "frac_sat": 0.006535000167787075 |
| }, |
| "enc_blocks.5.norm1": { |
| "mean": 0.00028194073820486665, |
| "std": 1.0010477304458618, |
| "abs_max": 9.194173812866211, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv1": { |
| "mean": -0.44563060998916626, |
| "std": 2.8831522464752197, |
| "abs_max": 17.75, |
| "frac_dead": 1.0172526344831567e-05, |
| "frac_sat": 0.00896453857421875 |
| }, |
| "enc_blocks.5.t_proj": { |
| "mean": -0.0013833555858582258, |
| "std": 0.10456022620201111, |
| "abs_max": 0.859375, |
| "frac_dead": 5.425347262644209e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.norm2": { |
| "mean": -0.010284450836479664, |
| "std": 0.9922280311584473, |
| "abs_max": 5.930111885070801, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "enc_blocks.5.conv2": { |
| "mean": 0.49203333258628845, |
| "std": 2.673565626144409, |
| "abs_max": 17.375, |
| "frac_dead": 5.086263172415784e-06, |
| "frac_sat": 0.00267791748046875 |
| }, |
| "enc_attns.5.norm": { |
| "mean": 0.0005484223365783691, |
| "std": 0.9975232481956482, |
| "abs_max": 8.068681716918945, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.qkv": { |
| "mean": 0.0058145360089838505, |
| "std": 0.7659077644348145, |
| "abs_max": 6.75, |
| "frac_dead": 6.668655987596139e-05, |
| "frac_sat": 0.0 |
| }, |
| "enc_attns.5.proj": { |
| "mean": -0.004640444181859493, |
| "std": 0.3258810043334961, |
| "abs_max": 4.125, |
| "frac_dead": 0.00016276042151730508, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm1": { |
| "mean": -4.7561195970047265e-05, |
| "std": 1.0003793239593506, |
| "abs_max": 7.752286434173584, |
| "frac_dead": 2.543131586207892e-06, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv1": { |
| "mean": -0.9362229108810425, |
| "std": 1.927423119544983, |
| "abs_max": 22.125, |
| "frac_dead": 1.4411079064302612e-05, |
| "frac_sat": 0.007679409347474575 |
| }, |
| "mid1.t_proj": { |
| "mean": 0.002691776491701603, |
| "std": 0.09969361871480942, |
| "abs_max": 0.72265625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.norm2": { |
| "mean": -0.012542976066470146, |
| "std": 0.9945948123931885, |
| "abs_max": 8.536553382873535, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid1.conv2": { |
| "mean": -0.007705233991146088, |
| "std": 0.5925304293632507, |
| "abs_max": 3.484375, |
| "frac_dead": 3.221299994038418e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.norm": { |
| "mean": 0.0004566659627016634, |
| "std": 0.997485876083374, |
| "abs_max": 7.793497085571289, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.qkv": { |
| "mean": 0.010610872879624367, |
| "std": 0.6843896508216858, |
| "abs_max": 5.8125, |
| "frac_dead": 6.527370715048164e-05, |
| "frac_sat": 0.0 |
| }, |
| "mid_attn.proj": { |
| "mean": -0.0158980805426836, |
| "std": 0.2606736719608307, |
| "abs_max": 1.4765625, |
| "frac_dead": 0.00015004475426394492, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm1": { |
| "mean": 0.0004634691576939076, |
| "std": 1.0005942583084106, |
| "abs_max": 7.7080841064453125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.conv1": { |
| "mean": -1.05628502368927, |
| "std": 1.9699069261550903, |
| "abs_max": 24.75, |
| "frac_dead": 1.1867947250721045e-05, |
| "frac_sat": 0.011021931655704975 |
| }, |
| "mid2.t_proj": { |
| "mean": 0.0014389199204742908, |
| "std": 0.10134941339492798, |
| "abs_max": 0.80859375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "mid2.norm2": { |
| "mean": -0.013441530987620354, |
| "std": 0.99652498960495, |
| "abs_max": 10.35103988647461, |
| "frac_dead": 0.0, |
| "frac_sat": 4.238552719471045e-06 |
| }, |
| "mid2.conv2": { |
| "mean": 0.01876220293343067, |
| "std": 0.8646733164787292, |
| "abs_max": 4.8125, |
| "frac_dead": 1.7801919966586865e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm1": { |
| "mean": 0.0012762596597895026, |
| "std": 1.001218557357788, |
| "abs_max": 8.005143165588379, |
| "frac_dead": 4.238552548940788e-07, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.conv1": { |
| "mean": -2.976426839828491, |
| "std": 4.75129508972168, |
| "abs_max": 59.75, |
| "frac_dead": 4.238552719471045e-06, |
| "frac_sat": 0.0577155239880085 |
| }, |
| "dec_blocks.0.t_proj": { |
| "mean": -0.0010795136913657188, |
| "std": 0.10378381609916687, |
| "abs_max": 0.6640625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.norm2": { |
| "mean": -0.017285624518990517, |
| "std": 0.9987004995346069, |
| "abs_max": 10.219768524169922, |
| "frac_dead": 8.477105097881577e-07, |
| "frac_sat": 3.3908420391526306e-06 |
| }, |
| "dec_blocks.0.conv2": { |
| "mean": 0.0011020468082278967, |
| "std": 0.4619784951210022, |
| "abs_max": 2.71875, |
| "frac_dead": 3.645155084086582e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.0.skip": { |
| "mean": 0.05044516175985336, |
| "std": 2.555394411087036, |
| "abs_max": 28.125, |
| "frac_dead": 1.9497341781971045e-05, |
| "frac_sat": 0.00231170654296875 |
| }, |
| "dec_attns.0.norm": { |
| "mean": 0.00031653710175305605, |
| "std": 0.9951615929603577, |
| "abs_max": 7.695018768310547, |
| "frac_dead": 1.6954210195763153e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.qkv": { |
| "mean": 0.005484588909894228, |
| "std": 0.659173846244812, |
| "abs_max": 6.28125, |
| "frac_dead": 8.024992712307721e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.0.proj": { |
| "mean": 0.009530746378004551, |
| "std": 0.541577935218811, |
| "abs_max": 4.25, |
| "frac_dead": 8.561876165913418e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.0.op": { |
| "mean": 0.13259148597717285, |
| "std": 5.4154438972473145, |
| "abs_max": 39.5, |
| "frac_dead": 4.026624992548022e-06, |
| "frac_sat": 0.07323244214057922 |
| }, |
| "dec_blocks.1.norm1": { |
| "mean": 0.0012307068100199103, |
| "std": 1.0010651350021362, |
| "abs_max": 9.00831127166748, |
| "frac_dead": 1.0596381798677612e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.conv1": { |
| "mean": -2.3529160022735596, |
| "std": 4.61397647857666, |
| "abs_max": 79.0, |
| "frac_dead": 8.053249985096045e-06, |
| "frac_sat": 0.0458882637321949 |
| }, |
| "dec_blocks.1.t_proj": { |
| "mean": -0.0017044859705492854, |
| "std": 0.10656358301639557, |
| "abs_max": 0.8046875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.norm2": { |
| "mean": -0.018469274044036865, |
| "std": 1.0006890296936035, |
| "abs_max": 15.932745933532715, |
| "frac_dead": 1.9073486328125e-06, |
| "frac_sat": 0.00025876364088617265 |
| }, |
| "dec_blocks.1.conv2": { |
| "mean": -0.001623636344447732, |
| "std": 0.4008197486400604, |
| "abs_max": 2.21875, |
| "frac_dead": 3.645155084086582e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.1.skip": { |
| "mean": 0.03524673357605934, |
| "std": 2.9326562881469727, |
| "abs_max": 18.125, |
| "frac_dead": 1.52587890625e-05, |
| "frac_sat": 0.0042321947403252125 |
| }, |
| "dec_attns.1.norm": { |
| "mean": 6.92961984896101e-05, |
| "std": 0.9948720932006836, |
| "abs_max": 7.577223300933838, |
| "frac_dead": 1.4834934063401306e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.qkv": { |
| "mean": 0.01385430246591568, |
| "std": 0.7135639786720276, |
| "abs_max": 5.875, |
| "frac_dead": 8.067378075793386e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.1.proj": { |
| "mean": 0.0175122432410717, |
| "std": 0.7606446146965027, |
| "abs_max": 4.0625, |
| "frac_dead": 5.955166489002295e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.1.op": { |
| "mean": 0.2810516059398651, |
| "std": 5.511878967285156, |
| "abs_max": 31.25, |
| "frac_dead": 4.132588856009534e-06, |
| "frac_sat": 0.08329842239618301 |
| }, |
| "dec_blocks.2.norm1": { |
| "mean": 0.0003542789781931788, |
| "std": 1.001099705696106, |
| "abs_max": 10.847487449645996, |
| "frac_dead": 7.770679530949565e-07, |
| "frac_sat": 1.059638137235197e-07 |
| }, |
| "dec_blocks.2.conv1": { |
| "mean": -1.4827497005462646, |
| "std": 4.093748092651367, |
| "abs_max": 52.75, |
| "frac_dead": 1.3033549294050317e-05, |
| "frac_sat": 0.05096329748630524 |
| }, |
| "dec_blocks.2.t_proj": { |
| "mean": -0.004567438270896673, |
| "std": 0.12303908169269562, |
| "abs_max": 0.79296875, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.norm2": { |
| "mean": -0.017762627452611923, |
| "std": 1.0039026737213135, |
| "abs_max": 12.032835960388184, |
| "frac_dead": 0.0, |
| "frac_sat": 1.3775295883533545e-05 |
| }, |
| "dec_blocks.2.conv2": { |
| "mean": 0.01796824112534523, |
| "std": 0.32834163308143616, |
| "abs_max": 2.265625, |
| "frac_dead": 5.955166489002295e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.2.skip": { |
| "mean": -0.02365983836352825, |
| "std": 3.8385989665985107, |
| "abs_max": 24.625, |
| "frac_dead": 1.229180270456709e-05, |
| "frac_sat": 0.020438512787222862 |
| }, |
| "dec_attns.2.norm": { |
| "mean": -0.00018475121760275215, |
| "std": 0.9939836263656616, |
| "abs_max": 6.334526538848877, |
| "frac_dead": 1.271565793103946e-06, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.qkv": { |
| "mean": -0.009542924351990223, |
| "std": 0.6096933484077454, |
| "abs_max": 4.90625, |
| "frac_dead": 0.00012051617522956803, |
| "frac_sat": 0.0 |
| }, |
| "dec_attns.2.proj": { |
| "mean": -0.033470239490270615, |
| "std": 0.44429004192352295, |
| "abs_max": 2.03125, |
| "frac_dead": 0.00014177958655636758, |
| "frac_sat": 0.0 |
| }, |
| "dec_ups.2.op": { |
| "mean": -0.3541581928730011, |
| "std": 4.1034979820251465, |
| "abs_max": 27.125, |
| "frac_dead": 7.920794814708643e-06, |
| "frac_sat": 0.03058568574488163 |
| }, |
| "dec_blocks.3.norm1": { |
| "mean": -0.0014417987549677491, |
| "std": 1.0000892877578735, |
| "abs_max": 10.300485610961914, |
| "frac_dead": 1.4967388324294006e-06, |
| "frac_sat": 3.973643103449831e-08 |
| }, |
| "dec_blocks.3.conv1": { |
| "mean": -0.9749405980110168, |
| "std": 2.3514747619628906, |
| "abs_max": 41.5, |
| "frac_dead": 1.3457404747896362e-05, |
| "frac_sat": 0.012159109115600586 |
| }, |
| "dec_blocks.3.t_proj": { |
| "mean": -0.0004592237819451839, |
| "std": 0.1721496880054474, |
| "abs_max": 1.2265625, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.norm2": { |
| "mean": -0.01709667034447193, |
| "std": 0.9970768094062805, |
| "abs_max": 14.025443077087402, |
| "frac_dead": 7.94728578057402e-07, |
| "frac_sat": 2.9987759262439795e-05 |
| }, |
| "dec_blocks.3.conv2": { |
| "mean": 0.015155193395912647, |
| "std": 0.4408976137638092, |
| "abs_max": 4.8125, |
| "frac_dead": 4.654460644815117e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.3.skip": { |
| "mean": 0.025708815082907677, |
| "std": 2.072154998779297, |
| "abs_max": 14.8125, |
| "frac_dead": 2.8742684662574902e-05, |
| "frac_sat": 0.0005844964180141687 |
| }, |
| "dec_ups.3.op": { |
| "mean": -0.49303382635116577, |
| "std": 2.7450525760650635, |
| "abs_max": 27.5, |
| "frac_dead": 1.2331538528087549e-05, |
| "frac_sat": 0.004969603382050991 |
| }, |
| "dec_blocks.4.norm1": { |
| "mean": -0.003529828740283847, |
| "std": 0.9992601871490479, |
| "abs_max": 15.095056533813477, |
| "frac_dead": 1.3598689747595927e-06, |
| "frac_sat": 2.0883701381535502e-06 |
| }, |
| "dec_blocks.4.conv1": { |
| "mean": -0.6139392256736755, |
| "std": 2.1271767616271973, |
| "abs_max": 26.375, |
| "frac_dead": 1.2238820090715308e-05, |
| "frac_sat": 0.0013043085345998406 |
| }, |
| "dec_blocks.4.t_proj": { |
| "mean": -0.005459681153297424, |
| "std": 0.1980716437101364, |
| "abs_max": 1.2578125, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.norm2": { |
| "mean": -0.008547251112759113, |
| "std": 0.9963138103485107, |
| "abs_max": 11.549351692199707, |
| "frac_dead": 2.2517311037972831e-07, |
| "frac_sat": 2.2517311037972831e-07 |
| }, |
| "dec_blocks.4.conv2": { |
| "mean": 0.03798123076558113, |
| "std": 0.6906255483627319, |
| "abs_max": 4.375, |
| "frac_dead": 4.7193632781272754e-05, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.4.skip": { |
| "mean": -0.16517576575279236, |
| "std": 1.960332989692688, |
| "abs_max": 19.75, |
| "frac_dead": 3.20937906508334e-05, |
| "frac_sat": 0.00011465284478617832 |
| }, |
| "dec_ups.4.op": { |
| "mean": -0.42157071828842163, |
| "std": 2.7456705570220947, |
| "abs_max": 41.5, |
| "frac_dead": 1.6805199265945703e-05, |
| "frac_sat": 0.013286677189171314 |
| }, |
| "dec_blocks.5.norm1": { |
| "mean": -0.00543779693543911, |
| "std": 0.997042715549469, |
| "abs_max": 11.520453453063965, |
| "frac_dead": 1.2483861837608856e-06, |
| "frac_sat": 6.291601550856285e-08 |
| }, |
| "dec_blocks.5.conv1": { |
| "mean": -0.04659652337431908, |
| "std": 1.7403225898742676, |
| "abs_max": 35.5, |
| "frac_dead": 1.5106465980352368e-05, |
| "frac_sat": 0.000550044933333993 |
| }, |
| "dec_blocks.5.t_proj": { |
| "mean": 0.015368163585662842, |
| "std": 0.27995017170906067, |
| "abs_max": 2.234375, |
| "frac_dead": 0.0, |
| "frac_sat": 0.0 |
| }, |
| "dec_blocks.5.norm2": { |
| "mean": 0.000441109761595726, |
| "std": 1.006426215171814, |
| "abs_max": 16.939395904541016, |
| "frac_dead": 4.635916752704361e-07, |
| "frac_sat": 5.116065494803479e-06 |
| }, |
| "dec_blocks.5.conv2": { |
| "mean": -1.157901406288147, |
| "std": 2.146827220916748, |
| "abs_max": 28.875, |
| "frac_dead": 2.6341942430008203e-05, |
| "frac_sat": 0.004018975887447596 |
| }, |
| "dec_blocks.5.skip": { |
| "mean": -0.401109904050827, |
| "std": 1.9778581857681274, |
| "abs_max": 26.875, |
| "frac_dead": 4.436572271515615e-05, |
| "frac_sat": 0.0012669430579990149 |
| }, |
| "out_norm": { |
| "mean": -0.006768459919840097, |
| "std": 0.9930412769317627, |
| "abs_max": 11.80453872680664, |
| "frac_dead": 1.4570024120530434e-07, |
| "frac_sat": 3.344482877309929e-07 |
| }, |
| "out_conv": { |
| "mean": 0.5060885548591614, |
| "std": 0.23508810997009277, |
| "abs_max": 1.21875, |
| "frac_dead": 2.966986812680261e-06, |
| "frac_sat": 0.0 |
| } |
| } |
| } |
