9?i ddlmZddlmZmZmZmZmZddlm Z ddl m Z ddl m Z edZeeedfeefZed d ed ed eeeeffd Zdeed ed eeeeffdZded efdZdedeed ed eefdZy)) lru_cache)ListUnionTypeVarTupleSequence) EinopsError) get_backend)ParsedExpressionTensor.)maxsizepatternopnamereturnc |j}t|}t|t|k7rtd|d|dd|vrtd|d|d|D]7}|dk7s t j |\}}|r$td|d|d|d|j d}t||z dz }||z} ||| fS) NzDuplicates in axes names in z(..., "z")*z No *-axis in zInvalid axis name z in )splitsetlenr r check_axis_name_return_reasonindex) rraxesaxes_setaxisis_validreason n_axes_before n_axes_aftermin_axess d/var/www/html/leadgen/airagagent/ocr_fallback/ocr_env/lib/python3.12/site-packages/einops/packing.pyanalyze_patternr#s ==?D4yH 4yCM!8yPRSTT (M& DEE] 3;/MMdS Hf!$6tfDPWyXZ"[\\ ] JJsOMt9},q0L|+H , 00tensorsc t|d\}}}t|d}g}g}t|D]\}} |j| } t | |krt d|d| d|d|d t | |z } |j | || |j |j| g| d|d | | d|j|| |fS) a Packs several tensors into one. See einops tutorial for introduction into packing (and how it replaces stack and concatenation). Parameters: tensors: tensors to be packed, can be of different dimensionality pattern: pattern that is shared for all inputs and output, e.g. "i j * k" or "batch seq *" Returns: (packed_tensor, packed_shapes aka PS) Example: ```python >>> from numpy import zeros as Z >>> inputs = [Z([2, 3, 5]), Z([2, 3, 7, 5]), Z([2, 3, 7, 9, 5])] >>> packed, ps = pack(inputs, 'i j * k') >>> packed.shape, ps ((2, 3, 71, 5), [(), (7,), (7, 9)]) ``` In this example, axes were matched to: i=2, j=3, k=5 based on order (first, second, and last). All other axes were 'packed' and concatenated. PS (packed shapes) contains information about axes that were matched to '*' in every input. Resulting tensor has as many elements as all inputs in total. Packing can be reversed with unpack, which additionally needs PS (packed shapes) to reconstruct order. ```python >>> inputs_unpacked = unpack(packed, ps, 'i j * k') >>> [x.shape for x in inputs_unpacked] [(2, 3, 5), (2, 3, 7, 5), (2, 3, 7, 9, 5)] ``` Read the tutorial for introduction and application scenarios. packrzpacked tensor #z' (enumeration starts with 0) has shape z, while pattern z assumes at least z axesN)r) r#r enumerateshaperr appendreshapeconcat) r%rrr r!backendreshaped_tensors packed_shapesitensorr*axis_after_packed_axess r"r'r'"sH->*> ? NNr$xc"d}|D]}||z} |S)Nr)r4resultr1s r"prodr8[s$ F !  Mr$r2r0ct|d\}}}t|}|j|}t||dz|zk7rt d|d||}|D cgc]} d| vrdn t | } } t d| D} | dkDrt d|d|d d gt|z||gz} | d k(r%t| d dD]\} }| | |z| | dz<nf| jd}t|D]} | | | | z| | dz<t|dzt| d d dD]}| |dz| |z | |<|d |}||dzd }td d f|z} t|D cgc]9\} }|j|g|t| | | | dzg|||;c}} Scc} wcc}} w#t$rtd |d | dd|wxYw)a8 Unpacks a single tensor into several by splitting over a selected axes. See einops tutorial for introduction into packing (and how it replaces stack and concatenation). Parameters: tensor: tensor to be unpacked packed_shapes: packed_shapes (aka PS) is a list of shapes that take place of '*' in each output. output will contain a single tensor for every provided shape pattern: pattern that is shared for input and all outputs, e.g. "i j * k" or "batch seq *", where * designates an axis to be unpacked Returns: list of tensors If framework supports views, results are views to the original tensor. Example: ```python >>> from numpy import zeros as Z >>> inputs = [Z([2, 3, 5]), Z([2, 3, 7, 5]), Z([2, 3, 7, 9, 5])] >>> packed, ps = pack(inputs, 'i j * k') >>> packed.shape, ps ((2, 3, 71, 5), [(), (7,), (7, 9)]) ``` In this example, axes were matched to: i=2, j=3, k=5 based on order (first, second, and last). All other axes were 'packed' and concatenated. PS (packed shapes) contains information about axes that were matched to '*' in every input. Resulting tensor has as many elements as all inputs in total. Packing can be reversed with unpack, which additionally needs PS (packed shapes) to reconstruct order. ```python >>> inputs_unpacked = unpack(packed, ps, 'i j * k') >>> [x.shape for x in inputs_unpacked] [(2, 3, 5), (2, 3, 7, 5), (2, 3, 7, 9, 5)] ``` Read the tutorial for introduction and application scenarios. unpack)rrz unpack(..., z)) received input of wrong dim with shape r(c38K|]}t|dk(yw)r(N)int).0r4s r" zunpack..s!Q1#a2g,!Qsz) received more than one -1 in z and can't infer dimensionsrNzError during unpack(..., "z!"): could not split axis of size z into requested )r#r r*rr r8sumr)rrangeslicer, Exception RuntimeError)r2r0rrr r!r. input_shape unpacked_axisp_shapelengths_of_composed_axesn_unknown_composed_axessplit_positionsr1r4unknown_composed_axisj shape_start shape_end slice_filler element_shapes r"r:r:bsR-%C%CB%G,- VA%4Q%7:RST:U%UOAE " V,q0#6N2OPQUSUQUV VA!0Q!7:RST:U!UOA  Vn}-KMA-/0I$%'-7L %.m$<  != OOYYu_Q-?QRUVQVAW'XYZ:+: : :   ;+p:   ( 1RSbceSfRg}o /   s$F0 F;.>F5,F;5F;;!GN) functoolsrtypingrrrrreinopsr einops._backendsr einops.parsingr r r<Shapestrr#r'r8r:r6r$r"rWs88(+   eCHotCy() 31S1#1%S# 2F11&6O(6"6OS6OU64;;N5O6OrEc\ 6\ $u+\ \ V \ r$