>:j6,SSKJr SSKr"SS5rg))OptionalNc\rSrSrSrSS\\S\\S\\S\\S\S \\S \S \\4S jjr S r Sr Sr S\ R4Sjr\S\\ R\ R\ R44Sj5r\SS\\ R\ R44Sjj5rSSjrSSjrS\ R4Sjr\SS\S\S\4Sjj5rSrg)IncrementalPCAa% An implementation of Incremental Principal Components Analysis (IPCA) that leverages PyTorch for GPU acceleration. Adapted from https://github.com/scikit-learn/scikit-learn/blob/main/sklearn/decomposition/_incremental_pca.py This class provides methods to fit the model on data incrementally in batches, and to transform new data based on the principal components learned during the fitting process. Args: n_components (int, optional): Number of components to keep. If `None`, it's set to the minimum of the number of samples and features. Defaults to None. copy (bool): If False, input data will be overwritten. Defaults to True. batch_size (int, optional): The number of samples to use for each batch. Only needed if self.fit is called. If `None`, it's inferred from the data and set to `5 * n_features`. Defaults to None. svd_driver (str, optional): name of the cuSOLVER method to be used for torch.linalg.svd. This keyword argument only works on CUDA inputs. Available options are: None, gesvd, gesvdj, and gesvda. Defaults to None. lowrank (bool, optional): Whether to use torch.svd_lowrank instead of torch.linalg.svd which can be faster. Defaults to False. lowrank_q (int, optional): For an adequate approximation of n_components, this parameter defaults to n_components * 2. lowrank_niter (int, optional): Number of subspace iterations to conduct for torch.svd_lowrank. Defaults to 4. lowrank_seed (int, optional): Seed for making results of torch.svd_lowrank reproducible. N n_componentscopy batch_size svd_driverlowrank lowrank_q lowrank_niter lowrank_seedc XlX lX0lX@lXPlX`lXplXlSUlUR(aUR5 gg)N) rrr r r r r r n_features__validate_lowrank_params) selfrrr r r r r rs W/home/mande/repo/quber/.venv/lib/python3.13/site-packages/peft/utils/incremental_pca.py__init__IncrementalPCA.__init__.sO) $$ "*( <<  ) ) + cURc-URc [S5eURS-UlgURUR:a [S5eg)NzKn_components must be specified when using lowrank mode with lowrank_q=None.z8lowrank_q must be greater than or equal to n_components.)r r ValueError)rs rr'IncrementalPCA._validate_lowrank_paramsGsY >> !  ( !noo!..2DN ^^d// /WX X0rcT[RRUSURS9$)NF) full_matricesdriver)torchlinalgsvdr rXs r _svd_fn_fullIncrementalPCA._svd_fn_fullOs"||tOOrcPURSLn[RRUS9 U(a [R"UR5 [R "XR URS9up4nX4UR4sSSS5 $!,(df  g=f)N)enabled)qniter) rrrandomfork_rng manual_seed svd_lowrankr r mH)rr" seed_enabledUSVs r_svd_fn_lowrankIncrementalPCA._svd_fn_lowrankRsu((4 \\ " "< " 8!!$"3"34''^^4CUCUVGA!: 9 8 8s A!B B%returncR[R[R/n[U[R5(d$[R "U[RS9nO!UR (aUR5nURup4URcOWURU:a[SURSUS35eURU:a[SURSU35eURU;aUR[R5nU$)z Validates and converts the input data `X` to the appropriate tensor format. Args: X (torch.Tensor): Input data. Returns: torch.Tensor: Converted to appropriate format. )dtypez n_components=z invalid for n_features=z<, need more rows than columns for IncrementalPCA processing.z6 must be less or equal to the batch number of samples ) rfloat32float64 isinstanceTensortensorrcloneshaperrr6to)rr" valid_dtypes n_samples n_featuress r_validate_dataIncrementalPCA._validate_dataZs u}}5 !U\\** Qemm4A YY A !    $     + 1 122J:,WMM    * 1 122hirhst  77, &U]]#ArcURSS:XaXU4$US:aUc [S5eUc [S5e[R"URS/URS9nX4-nUc<[R "URS[R URS9nOX-nURS[R S9nXg-U- nXt- n X - n U RS[R S9R5n U R5 U RS[R S9n XU- -n UcX- n O 0.z:last_variance should not be None if last_sample_count > 0.device)r6rF)dimr6) r=rrr;rFzerosr8sumsquaresquare_double)r" last_mean last_variancelast_sample_countnew_sample_countupdated_sample_countlast_sumnew_sum updated_meanTtemp correctionnew_unnormalized_varianceupdated_variancelast_unnormalized_variancelast_over_new_countupdated_unnormalized_variances r_incremental_mean_and_var(IncrementalPCA._incremental_mean_and_var}s 771:?->> > q   !YZZ$ !]^^ << QXXF0C  {{1771:U]]188TH 4H%%AU]]%3 *.BB  &uXX!5==X9@@B  $(HH%--H$H!!2B%BB!  8O )6)J &"3":":"! )) )o .2-K-K zz499d&:&:. *?  MA"ZZq1N A%*ZZ1E1E1L1L1NQ`1`dm0m%n "4 ^8STO ))..w7$:J:JJ#A <<++A.HA"((+HA"qu=T_q%89#$a4%))G4M*N#N .1 1 12 !"5D$5$5 6  #56I8I8I#J )ABUDDUDU)V&   Y$; ;#56G6G6I#J#O#O#QD  $)<<AHH#ED  rcXR- n[R"UR5URR 5R UR5$)aT Applies dimensionality reduction to `X`. The input data `X` is projected on the first principal components previously extracted from a training set. Args: X (torch.Tensor): New data tensor with shape (n_samples, n_features) to be transformed. Returns: torch.Tensor: Transformed data tensor with shape (n_samples, n_components). )r|rmmrMrzrVr>r6r!s r transformIncrementalPCA.transform-sA  Nxx D$4$4$6$67::177CCrnrrc## Sn[[X-55H nX1-nXR-U:aM[X55v UnM" X0:a[X05v gg7f)aGenerator to create slices containing `batch_size` elements from 0 to `n`. The last slice may contain less than `batch_size` elements, when `batch_size` does not divide `n`. Args: n (int): Size of the sequence. batch_size (int): Number of elements in each batch. min_batch_size (int, optional): Minimum number of elements in each batch. Defaults to 0. Yields: slice: A slice of `batch_size` elements. rN)rfintslice)rr rrstart_ends rrtIncrementalPCA.gen_batches<s^s1?+,A$C#a'# #E - 9/ ! sAA)r rzrrrr r r rr|rrr~rrr r})NTNNFNN)T)r)__name__ __module__ __qualname____firstlineno____doc__rrboolstrrrr#r2rr:rB staticmethodtupler^rnrwrurrt__static_attributes__rrrrsZ6'+#$($(#'&*,sm,tn,SM , SM ,  ,C=,,sm,2YP!5<<!F6D u||U\\5<<7 86D6Dp% ell8R2S2,EN Dell D"s""S""rr)typingrrrrrrrs ~"~"r