2Vh_:|ddlZddlZddlmZddlmZddlmZddlm Z ddl m Z edGdd e Z y) N)backend)ops) keras_export) TFDataLayer) tensorflowzkeras.layers.NormalizationcdeZdZdZ d fd ZdZdZdZdZdZ fdZ fd Z d Z d Z xZS) NormalizationaA preprocessing layer that normalizes continuous features. This layer will shift and scale inputs into a distribution centered around 0 with standard deviation 1. It accomplishes this by precomputing the mean and variance of the data, and calling `(input - mean) / sqrt(var)` at runtime. The mean and variance values for the layer must be either supplied on construction or learned via `adapt()`. `adapt()` will compute the mean and variance of the data and store them as the layer's weights. `adapt()` should be called before `fit()`, `evaluate()`, or `predict()`. Args: axis: Integer, tuple of integers, or None. The axis or axes that should have a separate mean and variance for each index in the shape. For example, if shape is `(None, 5)` and `axis=1`, the layer will track 5 separate mean and variance values for the last axis. If `axis` is set to `None`, the layer will normalize all elements in the input by a scalar mean and variance. When `-1`, the last axis of the input is assumed to be a feature dimension and is normalized per index. Note that in the specific case of batched scalar inputs where the only axis is the batch axis, the default will normalize each index in the batch separately. In this case, consider passing `axis=None`. Defaults to `-1`. mean: The mean value(s) to use during normalization. The passed value(s) will be broadcast to the shape of the kept axes above; if the value(s) cannot be broadcast, an error will be raised when this layer's `build()` method is called. variance: The variance value(s) to use during normalization. The passed value(s) will be broadcast to the shape of the kept axes above; if the value(s) cannot be broadcast, an error will be raised when this layer's `build()` method is called. invert: If `True`, this layer will apply the inverse transformation to its inputs: it would turn a normalized input back into its original form. Examples: Calculate a global mean and variance by analyzing the dataset in `adapt()`. >>> adapt_data = np.array([1., 2., 3., 4., 5.], dtype='float32') >>> input_data = np.array([1., 2., 3.], dtype='float32') >>> layer = keras.layers.Normalization(axis=None) >>> layer.adapt(adapt_data) >>> layer(input_data) array([-1.4142135, -0.70710677, 0.], dtype=float32) Calculate a mean and variance for each index on the last axis. >>> adapt_data = np.array([[0., 7., 4.], ... [2., 9., 6.], ... [0., 7., 4.], ... [2., 9., 6.]], dtype='float32') >>> input_data = np.array([[0., 7., 4.]], dtype='float32') >>> layer = keras.layers.Normalization(axis=-1) >>> layer.adapt(adapt_data) >>> layer(input_data) array([-1., -1., -1.], dtype=float32) Pass the mean and variance directly. >>> input_data = np.array([[1.], [2.], [3.]], dtype='float32') >>> layer = keras.layers.Normalization(mean=3., variance=2.) >>> layer(input_data) array([[-1.4142135 ], [-0.70710677], [ 0. ]], dtype=float32) Use the layer to de-normalize inputs (after adapting the layer). >>> adapt_data = np.array([[0., 7., 4.], ... [2., 9., 6.], ... [0., 7., 4.], ... [2., 9., 6.]], dtype='float32') >>> input_data = np.array([[1., 2., 3.]], dtype='float32') >>> layer = keras.layers.Normalization(axis=-1, invert=True) >>> layer.adapt(adapt_data) >>> layer(input_data) array([2., 10., 8.], dtype=float32) c t|di||d}nt|tr|f}n t |}||_||_||_||_d|_ d|_ d|_ |du|duk7rtd|d|y)NTzUWhen setting values directly, both `mean` and `variance` must be set. Received: mean=z and variance=) super__init__ isinstanceinttupleaxis input_meaninput_varianceinvertsupports_masking_build_input_shapemean ValueError)selfrrvariancerkwargs __class__s \/home/dcms/DCMS/lib/python3.12/site-packages/keras/src/layers/preprocessing/normalization.pyr zNormalization.__init__`s "6" <D c "7D;D & $"&   ($"6 7//3fN8*N  8c hyt_tfdjDrt ddjt t jDcgc]}|dk\r|n|zc}_jD]%}| t djdd|t fdtD_ tDcgc]}|jvrdnd c}_ tDcgc]}|jvr|nd c}_ t fd jD}|_ jgjd |d d _jd|dd _jdddd d _d_j'yt)j*j}t)j*j,}t)j.|j}t)j.|j}t)j0|j2_t)j0|j2_ycc}wcc}wcc}w)Nc38K|]}| kxs|k\ywNr ).0andims r z&Normalization.build..s#9!qD5y%AI%9szPAll `axis` values must be in the range [-ndim, ndim). Received inputs with ndim=z , while axis=rzDAll `axis` values to be kept must have a known shape. Received axis=z, inputs.shape=z, with unknown axis at index c3@K|]}|jvs|ywr!) _keep_axis)r"drs rr%z&Normalization.build..s!" at&>A" sc3(K|] }| ywr!r )r"r( input_shapes rr%z&Normalization.build..s"Ka;q>"KsrzerosF)nameshape initializer trainableronescountr r)r-r.dtyper/r0Tr3)lenranyrrrsortedr'range _reduce_axis_reduce_axis_mask_broadcast_shape_mean_and_var_shaper add_weight adapt_meanadapt_variancer2builtfinalize_staterconvert_to_tensorr broadcast_tocast compute_dtyperr)rr+r(mean_and_var_shaperrr$s`` @rbuildzNormalization.build{s   ;"- 9tyy9 9--1F- {L  tyyA!aAQX-A B  A1~% %%)YYK0$$/=12236 "" T{"   7?a4??2KN 9! #"K4??"KK#5 ?? ""oo(# .DO #'//(" #2#D # )DJDJ    !((9D,,T-@-@AH##D$*?*?@D''$2G2GHHT-?-?@DIHHXT5G5GHDMsB " ! s*J% 0J*J/ct|tjstj|r |j }nt|t jjr]t|jj }t|dk(r|jd}t|jj }|js|jnA|jD]2}||j |k7st#d|j d|t|tjrDtj$||j&}tj(||j&}ntj|rDt+j$||j&}t+j(||j&}nbt|t jjr&t+j,|j.}t+j,|j.}d}|D]tj0|j2t+j$|j&}t+j(|j&}|j&r+fd|j&D} t5j6| } nd} || z }t9| |z } d | z } || z|| zz} ||| z d zz| z||| z d zz| zz}| }nt;d t=||j>jA||jBjA||jEy ) aComputes the mean and variance of values in a dataset. Calling `adapt()` on a `Normalization` layer is an alternative to passing in `mean` and `variance` arguments during layer construction. A `Normalization` layer should always either be adapted over a dataset or passed `mean` and `variance`. During `adapt()`, the layer will compute a `mean` and `variance` separately for each position in each axis specified by the `axis` argument. To calculate a single `mean` and `variance` over the input data, simply pass `axis=None` to the layer. Arg: data: The data to train on. It can be passed either as a `tf.data.Dataset`, as a NumPy array, or as a backend-native eager tensor. If a dataset, *it must be batched*. Keras will assume that the data is batched, and if that assumption doesn't hold, the mean and variance may be incorrectly computed. r)z%The layer was built with input_shape=zO, but adapt() is being called with data with an incompatible shape, data.shape=)rrr4c3<K|]}j|ywr!)r.)r"r(batchs rr%z&Normalization.adapt..s*+, A*sg?zUnsupported data type: N)#rnpndarrayr is_tensorr.tfdataDatasetr element_specr5rKr@rGr'rrrr9varrr,r<rBrEmathprodfloatNotImplementedErrortyper>assignr?rA)rrQr+r( total_mean total_var total_count batch_mean batch_varbatch_reduce_shape batch_count batch_weightexisting_weightnew_total_meanrKs @radaptzNormalization.adapts* dBJJ '7+<+T%<%'>&?@==HMK  dBJJ 'D,=,=>Jt$*;*;->?J4+<+<=I bggoo .4#;#;"?K"#K{* $[1K? "% "40: 3LLn!< BB#$n!< BB '!! , 9 ,<&(?T |&LM M z* ""9- rc|j |jsytj|j|j |_tj|j |j|_tj|j|j |_ tj|j|j|_ yr!) rr@rreshaper>r;rrDrEr?rrs rrAzNormalization.finalize_states ?? &djj KK1F1FG HHTYY(:(:;  D$7$79N9NO 0B0BC rc |j td|jjj ||j }|j |j}|j |j}|jr|jjj||jjj||jjj|jjj|tjS|jjj|jjj!|||jjj|jjj|tjS)NzAYou must call `.build(input_shape)` on the layer before using it.r4)rrrcorerBrEconvert_weightrrnumpyaddmultiplymaximumsqrtepsilondividesubtract)rinputsrrs rcallzNormalization.call)sc 99 0 ""44 $,,5  ""499-&&t}}5 ;;<<%%)) ""++LL&&.. **//97??;L <<%%,, ""++FD9 ""**LL&&++H5w7H rc|Sr!r )rr+s rcompute_output_shapez"Normalization.compute_output_shapeLsrc$t|}|j|j|jt j |jjt j |jjd|S)N)rrrr) r get_configupdaterrrMarrayrtolistr)rconfigrs rryzNormalization.get_configOsj#%  ++188:HHT%8%89@@B    rcDt|||jyr!)r load_own_variablesrA)rstorers rrz Normalization.load_own_variables[s "5) rc8|jrd|jiSyNr+)rrhs rget_build_configzNormalization.get_build_config`s!  " "!4#:#:; ; #rc2|r|j|dyyr)rG)rr}s rbuild_from_configzNormalization.build_from_configds  JJvm, - r)NNF)__name__ __module__ __qualname____doc__r rGrerArurwryrrr __classcell__)rs@rr r sHPf9>6IIVVp D!F  <.rr )rUrlrM keras.srcrrkeras.src.api_exportr,keras.src.layers.preprocessing.tf_data_layerrkeras.src.utils.module_utilsrrPr r rrrs> -D9*+Y.KY.,Y.r