EiSSKrSSKJr SSKJrJr SSKrSSKJrJ r \ \S4r \\\ \ 4/\ \S44r/SQrS\S\4S jrS \R$R&S\\\44S jrS \\S\\\44S jrS\S\4Sjrg)N)Callable)AnyUnion)tree_flatten_with_pathtree_map.)normalize_source_namemodule_to_nested_dicttrack_dynamism_across_examplesclone_and_convert_to_metanamereturnc2[R"SSU5$)Nz\.([a-zA-Z_][a-zA-Z0-9_]*)z['\1'])resub)r s \/home/mande/repo/quber/.venv/lib/python3.13/site-packages/torch/fx/experimental/_dynamism.pyrrs 66/D AAmodulec0n0US'0US'[U5HnURS5(d[[X55(d|[X5n[ U[ R R5(dF[ U[[[ R45(a[U5[LaX1U'MMMMMM URSS9H upEXQSU'M URSS9H upFXaSU'M UR!5HupG[#U5USU'M U$![a GM-f=f)ziRecursively converts an nn.Module into a nested dictionary with explicit 'parameters' and 'modules' keys. _parameters_modules_F)recurse)dir startswithcallablegetattr isinstancetorchnnModuleintfloatTensortypeboolNotImplementedErrornamed_parameters named_buffersnamed_childrenr )r self_dict attr_name attr_valuer parambuffer submodules rr r sS "I!ImIj[  '',,X*66%V7 ":uxx??":UELL/IJJZ(4+5i(5K@ 6,!"..u.= ).- &>,,U,; )/- &<"002&;I&F *d#3 #   sB#D77 EEexample_inputsc0nUGHnSU;a=[US[RR5(a[ US5US'[ U5up4UGH)upV[U[ [[R45(dM2[U[R5(a[UR5nSnOU4nSnXQ;a2[[U55Vs/sH n[5PM snU4X'OWXupX:wa[U 5[U5:a3U R[55 [U 5[U5:aM3[U5HupXSU R!U 5 M GM, GM 0n UR#5HVunup[SU 55nSSR%SU55-nUSR&nUU ;a0U U'XUU'MX U $s snf) a This function analyzes a list of example inputs to determine the dynamism of their shapes. It tracks whether the dimensions of tensors or non-tensor values change across different examples. The function returns a dictionary where each key represents a path to a value in the input examples, and the corresponding value is a tuple indicating which dimensions are dynamic (i.e., change across examples). This helps in understanding how the structure of data varies across different instances. selfTFrc3># UHn[U5S:v M g7f)N)len).0ss r 1track_dynamism_across_examples..cs7h#a&1*hsLc3:# UHn[U5v M g7f)N)str)r6ks rr8r9ds>X3q6( Xs)rrrr r rr!r"r#tupleshaperanger5setappend enumerateadditemsjoinkey)r0trackingexleaves_with_pathsrkey_pathvaluer@ is_tensordim_setsflagidimoutput _is_tensor final_dynkey_strrHs rr r <s<>H RX>>>qkoo f F3K(s G -= M%'Is$G< example_inputc:S[S[4Sjn[X5$)z This function takes a list of example inputs and for each tensor, clones it and converts it to device=meta. For non-tensor values, it keeps the reference. It uses pytree to handle nested structures recursively. rMr c~[U[R5(aUR5R SS9$U$)Nmeta)device)rrr#cloneto)rMs r transform_fn/clone_and_convert_to_meta..transform_fnrs2 eU\\ * *;;=##6#2 2 r)rr)rWr^s rr r ls" CC L 00r)rcollections.abcrtypingrrrtorch.utils._pytreerrr?KeyPathr!r"NonTensorShapeFn__all__r=rrr dictr listr r rrris $ @ S/U3:./sCx@A BBB %((// d38n F-I- #s(^-` 1S 1S 1r