tj ddlZddlmZddlmZddlZddlmZmZe edfZ ee e zge edffZ gdZdedefd Zd ejjdeeeffd Zd eedeeeffd ZdedefdZdS)N)Callable)Any)tree_flatten_with_pathtree_map.)normalize_source_namemodule_to_nested_dicttrack_dynamism_across_examplesclone_and_convert_to_metanamereturnc.tjdd|S)Nz\.([a-zA-Z_][a-zA-Z0-9_]*)z['\1'])resub)r s ]/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/torch/fx/experimental/_dynamism.pyrrs 6/D A AAmoduleci}i}i}||d<||d<t|D]} |dstt||sqt||}t |t jjsBt |ttt j frt|tur|||<#t$rYwxYw|dD]\}}||d|<|dD]\}}||d|<|D]\}} t#| |d|<|S)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 parametersmodules attr_name attr_valuer parambuffer submodules rrrs "I*,J)+G)Im#Ij[[  '',, 6X **66 6%VY77 ":ux??6":UEL/IJJ6Z((44+5Ii("    H  ..u.==// e). - &&,,U,;;00 f)/ - &&!0022GGi&;I&F&F *d## sB#C CCexample_inputscBi}|D]}d|vr=t|dtjjrt |d|d<t |\}}|D]C\}}t|t ttjfs.t|tjrt|j }d}n|f}d}||vr,dtt|D|f||<ns||\} } | |kr t| t|krA| tt| t|kAt|D],\} } ||d| | -Ei} |D]d\}\} }td| D}ddd |Dz}|dj}|| vri| |<|| ||<e| S) 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. selfTFc*g|]}tS)set).0rs r z2track_dynamism_across_examples..Ys&H&H&Hsuu&H&H&Hrrc3<K|]}t|dkVdS)N)len)r7ss r z1track_dynamism_across_examples..es,77#a&&1*777777rLc36K|]}t|VdS)N)str)r7ks rr=z1track_dynamism_across_examples..fs*>>3q66 >>>>>>r)rrrrrrr r!r"tupleshaperanger;appendr6 enumerateadditemsjoinkey)r1trackingexleaves_with_pathsrkey_pathvaluerD is_tensordim_setsflagidimoutput _is_tensor final_dynkey_strrKs rr r >sL<>H22 R<>X>>>>>>qko f  F3K(s G Mr example_inputcFdtdtfd}t||S)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. rPr ct|tjr(|dS|S)Nmeta)device)rrr"cloneto)rPs r transform_fnz/clone_and_convert_to_meta..transform_fnts: eU\ * * 3;;==##6#22 2 r)rr)rZras rr r ns6 CC L- 0 00r)rcollections.abcrtypingrrtorch.utils._pytreerrrCKeyPathr r!NonTensorShapeFn__all__rArrrdictrlistr r r5rrrjsE $$$$$$ @@@@@@@@ S/S5[M5c?:;   BBBBBB "%(/"d38n""""J-I- #s(^----` 1S 1S 1 1 1 1 1 1r