+ j1+^RIt^RIHt^RIHt^RIHtHt^RIHtH t H t H t ^RI H t ^RIHt!RR ]4t]P!R R 44tR R ltRRltRtRtR#)NCallable) dataclass)AnyProtocol)_C_opsautogradTensor)_pytree)utilsc,a]tRt^ toV3RltRtVtR#) InfoProtocolcN<V^8dQh/S[R,;R&S[R,;R&#N _backward_fn_setup_context_fnr)format __classdict__s"X/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torch/_library/autograd.py __annotate__InfoProtocol.__annotate__ s#T/!$&N__name__ __module__ __qualname____firstlineno____annotate_func____static_attributes____classdictcell__rs@rrr rrc,a]tRt^toV3RltRtVtR#)InfocN<V^8dQh/S[R,;R&S[R,;R&#rr)rrs"rrInfo.__annotate__s%T/! $&rrNrr$s@rr'r'r%rr'cPV^8dQhR\PR\R\/#)ropinforeturn)r OpOverloadrr)rs"rrrs*~~4??~,~8~rc aaaaa a a a a RSP RSP RSP 2pSPp\P !S4o \P !SP4o \RVP44o \;QJd4R.VPOVPO4F 'gK RM) RM%!R.VPOVPO44o \!RR44oV3Rlo V VV3R lpV VV V3R lp\V\P3R \V4R \V4/4oS 'd \!S4oVVV VV 3R lpV#)GeneratedBackwardFor__c3B"TFqP'*xK R#5iN) kwarg_only.0as& r %make_autograd_impl..sI8H1,,..8Hsc3b"TF%p\P!VP4xK' R#5ir3)r is_tensorlist_like_typetyper5s& rr8r9s)#5A %%aff--5s-/TFc,a]tRt^$toV3RltRtVtR#)$make_autograd_impl..Metadatac\<V^8dQh/S[P;R&S[S[S[3,;R&#)rkeysetkeyword_only_args)rDispatchKeySetdictstrr)rrs"rr1make_autograd_impl..Metadata.__annotate__$s+!!! S>)rrNrr$s@rMetadatar>$r%rrFc<\P!4;_uu_4SP!V\P,.VO5/VBpVuuRRR4# +'giR#;ir3)r_AutoDispatchBelowAutograd redispatch_after_autograd_keyset)r@argskwargsresultr+s&&& rredispatch_no_grad.make_autograd_impl..redispatch_no_grad)sH  * * , ,]]6B,E,E#EWWPVWF- , , ,s -A A( c <VR,pVRRp\P!4;_uu_4VPpVPpSP!V\P ,.VO5/VBpSP 'dS\P!SPW4wrS'dSP WWER7MSP WVR7VuuRRR4# +'giR#;i)N)ctxinputskeyword_only_inputsoutput)rRrSrU) rrHr@rArIrJrr fill_defaults_schema) rRrKmetadatar@rLrMhas_kwarg_only_argsr,r+s &* rforward#make_autograd_impl..forward.s8CRy  * * , ,__F//F]]6B,E,E#EWWPVWF%%% %222::tL &**&+**s*O/- , , ,s B)C!! C2 c <S P'EdVPpVRR VnS P!V.VO5!pW n\V4^, pVS 0p\V\4'd \V4M^pWe9d/VS 8Xd \ V4MV RS 2p\ RS RV RV R24h\V\4'dbW4Rp\;QJdRV4F 'gK RM RM !RV44'd\ RS R 24hS 'dVRVp.VORN5#VR3#\ R S R 24h XTni;i) Nz or zThe backward formula for z5 returned an incorrect number of gradients (expected z, got z). Expected one gradient for each forward input, or for each positional input to the operator. Use None for inputs that do not require a gradient.c3("TFqRJxK R#5ir3r)r6grads& rr87make_autograd_impl..backward..es@KD4'KsTFz returned a non-None gradient for an input that was not passed to the operator. Defaulted inputs that were not passed through autograd must return None.zTrying to backward through zQ but no autograd formula was registered. Please use register_autograd to add one.rV)rneeds_input_gradlen isinstancetuplerD RuntimeErrorany) rRgradsprev_needs_input_gradrMnum_actual_inputsvalid_return_countsactualexpected extra_gradshas_tensorlist_like_argsr,num_positional_argsr+s &* rbackward$make_autograd_impl..backwardKs     =(+(<(<%' #46I"J $.vu$=$=S[1F0),??)*-.d3F2GH #/t455=JfVHM=>&%(($%78 3@K@333@K@@@&3B48=> ,#$6%67F&&4< )".7 8  =(=$s *D== Er[rpc~<S'dY\P!4'd5\P!V/VB'd\SP R24hS!WV4#\P!4'd8\P!V!'dSP !.VOS!W4N5!pV#S!WV4pV#)zu(): functions with out=... arguments don't support automatic differentiation, but one of the arguments requires grad.)ris_grad_enabled_any_requires_gradre_opnameapply) r@rKrArM GeneratedrF is_out_opr+rNs &*, r autograd_impl)make_autograd_impl..autograd_impls !!##(=(=)*))#zzl#%% &f4EF F    B$9$94$@$@__PdPHV,OPF (6GHF r) _namespaceru _overloadnamerXr is_outrZsum argumentsrfreturnsrr<r Function staticmethodsupports_tensorlist)r+r,nameschemar[rpryrwrFrZrnrxrorNsff @@@@@@@rmake_autograd_implrsF' a |1REUEUDVWD ZZF R I33BJJ?I8H8HII"s#56##5fnn5#sss#56##5fnn5#  *** :) ) V    |G,  X. I ' 2 $ rc0V^8dQhR\R\/#)rclsr-)r)rs"rrrsjjSjSjrcaaaaVPoVPoVPo\!RR44oVV3RlpV3RlpVV3RlpWnW nW0nV#)a$Allows a given autograd.Function class to support List[Tensor] inputs/outputs. Regular autograd.Function has a constraint that it only directly supports autograd for Tensors. Applying @supports_tensorlist enables an autograd.Function to support autograd for List[Tensor] inputs and outputs. c4a]tRt^toRtRtV3RltRtVtR#)/supports_tensorlist..TensorListMetadataNc<V^8dQh/S[P;R&S[PR,;R&S[R,;R&#)r input_specN output_specresult_is_tuple)r TreeSpecbool)rrs"rr.TensorListMetadata.__annotate__s?$$$%%,3+ rr) rrrr rrr!r"r#r$s@rTensorListMetadatars04 '+ rrc<VR,pVRRp\VS4'g \R4h\P!\ V4VP 4pS!V.VO5!p\V\ 4VnVP'gV3p\P!V\4wrEWRn \VR4'd \R4hW n \ V4#)rQNzNYI: calling supports_tensorlist autograd.Function.forward directly. You should probably be calling .apply instead. Please file an issue if not. _pt_metadataz@Please don't set ctx._pt_metadata; PyTorch uses it to store inforV)rcNotImplementedErrorr tree_unflattenlistrrdr tree_flattennot_list_of_tensorrhasattrrer)rRrKrYrM flat_resultrr orig_forwards&* r new_forward(supports_tensorlist..new_forwards8CRy($677%/  %%d4j(2E2EFc)D)#-fe#< '''YF#*#7#7@R#S * 3 ' 'R $[!!rc\<\VR4'g \R4hVPp\P!\ V4VP 4pVPp\P!\ VPRR4VP4VnS!V.VO5!pW0n\V\4'gV3p\P!V\4wrVWbP8wd\RV RVP R24h\VR.,4# Y0ni;i)rzNYI: calling supports_tensorlist autograd.Function.backward directly. This will automatically get called by PyTorch autograd. Please file an issue if you need this.NzRExpected the return from backward to be of the same structure as the inputs. Got: z (return from backward), z (inputs)rV)rrrr rrrrarrcrdrnot_list_of_optional_tensorre)rRrgrYrh grad_inputsflat_grad_inputsgrad_inputs_spec orig_backwards&* r new_backward)supports_tensorlist..new_backwards0sN++%9  ##&&tE{H4H4HI!$ 4 4 9#*#9#9S))#2./1D1D$C (4e4K#8 +u--&.K .5-A-A 4. * 22 2''7&88Q&&'y2  %.//%$9 s "AD##D+c<\P!V\R7wrS!V4pS!.VOVN5!pVPf \ R4h\P !\ V4VP4pVP'g^\V\4'g\ R\V4 24h\V4^8wd\ R\V4 24hV^,#V#))is_leafz%metadata.output_spec must not be Nonezresult must be tuple, got z%result tuple must have length 1, got ) r rrrAssertionErrorrrrrcrdr<rb)rK flat_argsrrYrMr orig_applys* r new_apply&supports_tensorlist..new_applys ' 4 4TCU V %j11Y11    ' !HI I''V h6J6JK'''fe,,$'A$v,%PQQ6{a$;CK=I!9  r)r[rprvr)rrrrrrrrs& @@@@rrrs];;LLLMJ,,, "2*0X"KLI Jrc\V\4'dR#\V\4'd5\;QJdRV4F 'gK R# R#!RV44#R#)Fc3L"TFp\V\4'*xK R#5ir3rcr r6ls& rr8%not_list_of_tensor.. s;dz!V,,,ds"$Trcrdrrftrees&rrrsK$$s;d;ss;s;s;d;;; rc\V\4'dR#\V\4'd5\;QJdRV4F 'gK R# R#!RV44#R#)Fc3b"TF%qRJ;'d\V\4'*xK' R#5ir3rrs& rr8.not_list_of_optional_tensor..s'M1D=>>Av)>%>>s//Trrs&rrrsK$$sMMssMsMsMMMM r) dataclassescollections.abcrrtypingrrtorchrr r r torch.utilsr r rr'rrrrrrrrs`$! ,,'8'  ''' ~BjZr