RЦi0 :SSKrSSKrSSKrSSKJr SSKJr SSKJrJ r J r J r J r J r JrJrJr SSKrSSKrSSKJr SSKJr SSKJr /SQrS.S \R0S \4S jjrS/S \ S \R0S\S\S\R04 SjjrS/S \R0S\S\S\4SjjrS/S \R0S\S\S\4SjjrSrS0S\ \\\4S\\ \ 4S\S\4Sjjr S0S \R0S\S\4Sjjr!S0S \R0S\S\4Sjjr"S1S\ \\\R04S\#S\\\\S 44S!\\\\\R044S"jjr$SS#.S$\ \4S%jjr%S2S&\#S'\S(\S$\ \4S)jjr&S*\#S+\S\4S,jr'S \R0S\ \4S-jr(g)3N) defaultdict)chain) AnyCallableDictIteratorListOptionalTupleTypeUnion)nn)Tensor)use_reentrant_ckpt) model_parameters named_apply named_modulesnamed_modules_with_paramsadapt_input_convgroup_with_matcher group_modulesgroup_parametersflatten_modulescheckpoint_seq checkpointreinit_non_persistent_buffersmodel exclude_headcU(a#UR5Vs/sHo"PM snSS$UR5$s snf)N) parameters)rrps V/var/www/html/ai-image-ml/venv/lib/python3.13/site-packages/timm/models/_manipulate.pyrrs@ ++-.-a-.s33!!/s ;fnmodule depth_first include_rootreturnc U(d U(aU"XS9 UR5H+upVU(aSRX%45OUn[XXSSS9 M- U(a U(aU"XS9 U$)N)r%name.T)r$r%r*r&r')named_childrenjoinr)r$r%r*r&r' child_name child_modules r#rrs] < &$$*$9$9$; 59SXXt01z rZgkl%<| &$ Mr*c## U(d U(aX4v UR5H3upEU(aSRX45OUn[XTUSS9ShvN M5 U(aU(aX4v gggN7fNr+T)r%r*r&r')r,r-rr%r*r&r'r.r/s r#rr-sx <l$*$9$9$; 59SXXt01z  kX\^ ^ ^%<|l${ ^sAA5A3A5c#4# UR(aU(d U(aX4v UR5H3upEU(aSRX45OUn[XTUSS9ShvN M5 UR(aU(aU(aX4v ggggN07fr2) _parametersr,r-rr3s r#rr=s +,l$*$9$9$; 59SXXt01z ,kX\^ ^ ^%<kll/;k ^sA#B%B&1B)i named_objects group_matcher return_valuesreversec:^[T[5(a/n[TR55HuunupgUcM [U[[ 45(a/UH'nU[ R"US5U4US4/- nM) MWU[ R"U5U4S4/- nMw UmU4Sjn [[ 5n UH'upX"U 5RU(aU OU 5 M) [[ 5n Sn[[SU R555H3n US:dU S[S:waUS- nXRX5 M5 U(a9U(aS5e0nU R5HunnUHnXU'M M U$U $)Nrc >[T[[45(a{THiupnURU5nU(dM X$R 5U4n[[ [ [R"[SU5555s $ [ S54$T"U5n[U[RR5(dU4$[U5$)Ninf) isinstancelisttuplematchgroupsmapfloatr from_iterablefilter collectionsabcIterable)r*match_fnprefixsuffixrpartsordr7s r# _get_grouping)group_with_matcher.._get_groupinges mdE] 3 3,9(&NN4(1#XXZ8E UE,?,?tU@S,T!UVV -: <= %Cc;??#;#;<<t : r0c USL$)N)xs r#$group_with_matcher..|sQd]r0z-reverse mapping only sensible for name output)r>dict enumerateitemsr@r?recompilerappendsortedrFkeysMATCH_PREV_GROUPextend)r6r7r8r9compiled group_ordinal group_namemspecsspecrPgroupingkvlayer_id_to_paramlidparam_to_layer_idlmns ` r#rrPs -&&2;Mr@strr Sequential ModuleList ModuleDictrr,r-)rrtrKruprefix_is_tupler*r%s r#rrs !/O,$$ ; &MM2=="--HLMM+L%  Z55&%%' "1wt)   'l"88VN3Dl"& sB5C67C48=C6 use_reentrantr~c|Uc [5n[RRR"U/UQ7SU0UD6$)zqcheckpoint wrapper fn A thin wrapper around torch.utils.checkpoint.checkpoint to default use_reentrant to False r~)rtorchutilsr)functionr~argskwargss r#rrsN*, ;; ! ! , ,   $    r0everyflatten skip_lastcJUc [5nSn[U[RR5(aUR 5nU(a[ R"U5n[U[[45(d [U5n[U5nU(aUS-nSn[SXr5HGn [X-S- US- 5n[RRRU"XU5UUS9nMI U(aU"US-[U5S- U5"U5$U$)aA helper function for checkpointing sequential models. Sequential models execute a list of modules/functions in order (sequentially). Therefore, we can divide such a sequence into segments and checkpoint each segment. All segments except run in :func:`torch.no_grad` manner, i.e., not storing the intermediate activations. The inputs of each checkpointed segment will be saved for re-running the segment in the backward pass. See :func:`~torch.utils.checkpoint.checkpoint` on how checkpointing works. .. warning:: Checkpointing currently only supports :func:`torch.autograd.backward` and only if its `inputs` argument is not passed. :func:`torch.autograd.grad` is not supported. .. warning: At least one of the inputs needs to have :code:`requires_grad=True` if grads are needed for model inputs, otherwise the checkpointed part of the model won't have gradients. Args: functions: A :class:`torch.nn.Sequential` or the list of modules or functions to run sequentially. x: A Tensor that is input to :attr:`functions` every: checkpoint every-n functions (default: 1) flatten: flatten nn.Sequential of nn.Sequentials skip_last: skip checkpointing the last function in the sequence if True use_reentrant: Use re-entrant checkpointing Returns: Output of running :attr:`functions` sequentially on :attr:`*inputs` Example: >>> model = nn.Sequential(...) >>> input_var = checkpoint_seq(model, input_var, every=2) c^^^UUU4SjnU$)NcJ>[TTS-5HnTU"U5nM U$)Nr;)range)_xjend functionsstarts r#forward5checkpoint_seq..run_function..forwards+5#'*q\"%+Ir0rT)rrrrs``` r# run_function$checkpoint_seq..run_functions r0r;rRrr})rr>rrrychildrenrrEr@r?lenrminrr) rrUrrrr~rnum_checkpointedrrs r#rrsV*, )UXX0011&&( '' 2 i% / /)$ 9~A Cq*2%-!#%5%9: KK " " - - Y / ' . 3C!GS^a%7CAFF Hr0in_chans conv_weightcURnUR5nURup4pVUS:XaVUS:a?URSS-S:XdeURX4S-SXV5nUR SSS9nO~UR SSS9nOmUS:wagUS:wa [ S5e[ [R"US- 55nURSUSS5SS2SU2SS2SS24nUS[U5- -nURU5nU$) Nr;rF)dimkeepdimTz*Weight format not supported by conversion.) dtyperDshapereshapesumNotImplementedErrorintmathceilrepeatto)rr conv_typeOIJKrs r#rr!s!!I##%K""JA!1} q5$$Q'!+q0 00%--aaAAK%//a/?K%//a/>K Q 6%&RS S8a<01F%,,Q1=a(Aq>PQK Ah/ 0K..+K r0c/nUR5HBup#[US5(dMUR5 URU(aUOS5 MD U$)aWalk model and call init_non_persistent_buffers() on modules that have it. This reinitializes computed buffers (like RoPE frequencies, attention bias indices) that are marked as non-persistent and thus not saved in checkpoints. These buffers are typically computed from module configuration and need to be reinitialized after loading a checkpoint. Args: model: Model to reinitialize buffers for Returns: List of module names that were reinitialized Example: >>> model = create_model('vit_base', pretrained=True) >>> # After loading checkpoint or moving to new device >>> reinitialized = reinit_non_persistent_buffers(model) >>> print(f"Reinitialized {len(reinitialized)} modules") init_non_persistent_buffersz(root))rhasattrrr])r reinitializedr*r%s r#rr:sR(M++-  68 9 9  . . 0  8 <. r0)F)TF)FF)r;r sequential)r;FFN))collections.abcrGrr[r itertoolsrtypingrrrrr r r r r rtorch.utils.checkpointrr timm.layersr__all__Moduleboolrrrxrrr`rrrrrrrrrrTr0r#rs #TTT * , "BII"T"!# "        YY $ "      $ "        $ :c3h0:T8^,:: :@$ ` `` `$ h hh h.0;G #c299n 56##c5c?*+#CtBII!778 #B%)D>2(, I I  I  I  ~ I XsF2tCyr0