,j FSSKrSSKJr SSKJrJrJrJr SSK r SSK J r SSK J r SSK J r 0S4SjrS r\"S 5r\"S 5r\"S 5r\"S 5rSrSS/S4SjrSrSS\S\\\\\4S\S\\\\44SjjrS\\\\S44S\\S44Sjrg)N)repeat)ListOptionalTupleUnion)nn)_assert)FrozenBatchNorm2dcUnSnU(aX!;nU(Ga[U[RRR[RRR 45(Ga'[ UR5nURUlURUlUR(aURRR5R5URl URRR5R5URl URRURl URRURl UR UlU$UR#5HCupVU(aSR%X%/5OUn['XaU5nXLdM2UR)XX5 ME U$)a Converts all `BatchNorm2d` and `SyncBatchNorm` layers of provided module into `FrozenBatchNorm2d`. If `module` is itself an instance of either `BatchNorm2d` or `SyncBatchNorm`, it is converted into `FrozenBatchNorm2d` and returned. Otherwise, the module is walked recursively and submodules are converted in place. Args: module (torch.nn.Module): Any PyTorch module. module_match (dict): Dictionary of full module names to freeze (all if empty) name (str): Full module name (prefix) Returns: torch.nn.Module: Resulting module Inspired by https://github.com/pytorch/pytorch/blob/a5895f85be0f10212791145bfedc0261d364f103/torch/nn/modules/batchnorm.py#L762 T.) isinstancermodules batchnorm BatchNorm2d SyncBatchNormr num_featuresaffineweightdataclonedetachbias running_mean running_varepsnamed_childrenjoinfreeze_batch_norm_2d add_module) module module_matchnameresis_match child_namechildfull_child_name new_childs ڝC:\Users\johnn\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.13_qbz5n2kfra8p0\LocalCache\local-packages\Python313\site-packages\open_clip/utils.pyrr sa CH'Jv (<(<(H(H"**J^J^JlJl'mnn 3 34!..]] ==$mm00668??ACJJO"KK,,224;;=CHHM & 3 3 8 8%1166** J "(!6!6!8 J>Bchh'9: O,U/RI%z5 "9 Jc^U4SjnU$)Nc>[U[RR5(aU$[ [ UT55$N)r collectionsabcIterabletupler)xns r*parse_ntuple..parse4s0 a11 2 2HVAq\""r+)r4r5s` r*_ntupler83s# Lr+c$[U5"U5$r.)r8)r4r3s r*r>?s Ar+c_fcc_projTcUR5GHXupE[[UR555S:a [ XQX#5 [ U[ RR5(dMcXB;dMjURUnU"URURURSL5URU'U(dMURURRRURR5 URURcGMURURRRUR5 GM[ U$Nr)rlenlistchildrenreplace_linearrtorchrLinear_modules in_features out_featuresrrrcopy_)modellinear_replacementinclude_modules copy_weightsr#r! old_modules r*rFrFCs ,,.  tFOO%& '! + 6 U fehhoo . .43J-J#5""## 4'$ENN4 |t$++0066z7H7H7M7MN>>$',,8NN4(--2288I/ Lr+cUR5HBn[US5(dMURRnUR 5 X!lMD g)Nprepare_for_eval)rhasattrrdtyperSint8_original_dtype)rMmrVs r*$convert_int8_model_to_inference_moderXVs@ ]]_ 1( ) )"#((..    $7 ! r+rindicesas_setreturnc  UcUn[U[5(aG[SUs=:=(a U:*Os SUSUS35 [U5Vs/sH o0U- U-PM nnOQ/nUHInUS:aX-OUn[SUs=:*=(a U:Os SUSUS- S35 UR U5 MK [ R R5(dU(a[U5[U54$U[U54$s snf)aDetermine the absolute feature indices to 'take' from. Note: This function can be called in forward() so must be torchscript compatible, which requires some incomplete typing and workaround hacks. Args: num_features: total number of features to select from indices: indices to select, None -> select all int -> select last n list/tuple of int -> return specified (-ve indices specify from end) as_set: return as a set Returns: List (or set) of absolute (from beginning) indices, Maximum index rzlast-n (z) is out of range (1 to )zfeature index z is out of range (0 to r9) rintr rangeappendrGjit is_scriptingsetmax)rrYrZi take_indicesidxs r*feature_take_indicesrh^s*'3G++|+xy@XYeXffg-hirus// 2/1r$P AJ AJ AJ AJ & @Fx>P_c&848&+&+%T#Y/0&+&+ 49c> &+RU3c3h#78U38_r+