6bi}6SrSSKrSSKJs Jr SSKJr SSKJ r SSK J r SSK J r SSKJr SSKJr \"S 5"S S 55r\"S S 5"SS\55r\"SS5"SS\55r\"SS5"SS\55r\"SS5"SS\55r\"SS5"SS\55r\r\r\r\r\r\r\r\r\"S 5S&S!j5r \"S"5S'S#j5r!\"S$5S%5r"g)(z@Constraints: functions that impose constraints on weight values.N)backend) serialization)deserialize_keras_object)serialize_keras_object) keras_export) doc_controlszkeras.constraints.Constraintc4\rSrSrSrSrSr\S5rSr g) Constraint!a+Base class for weight constraints. A `Constraint` instance works like a stateless function. Users who subclass this class should override the `__call__` method, which takes a single weight parameter and return a projected version of that parameter (e.g. normalized or clipped). Constraints can be used with various Keras layers via the `kernel_constraint` or `bias_constraint` arguments. Here's a simple example of a non-negative weight constraint: >>> class NonNegative(tf.keras.constraints.Constraint): ... ... def __call__(self, w): ... return w * tf.cast(tf.math.greater_equal(w, 0.), w.dtype) >>> weight = tf.constant((-1.0, 1.0)) >>> NonNegative()(weight) >>> tf.keras.layers.Dense(4, kernel_constraint=NonNegative()) cU$)a'Applies the constraint to the input weight variable. By default, the inputs weight variable is not modified. Users should override this method to implement their own projection function. Args: w: Input weight variable. Returns: Projected variable (by default, returns unmodified inputs). selfws R/home/james-whalen/.local/lib/python3.13/site-packages/tf_keras/src/constraints.py__call__Constraint.__call__;s c0$)zReturns a Python dict of the object config. A constraint config is a Python dictionary (JSON-serializable) that can be used to reinstantiate the same object. Returns: Python dict containing the configuration of the constraint object. r rs r get_configConstraint.get_configJs  rcU"S0UD6$)a:Instantiates a weight constraint from a configuration dictionary. Example: ```python constraint = UnitNorm() config = constraint.get_config() constraint = UnitNorm.from_config(config) ``` Args: config: A Python dictionary, the output of `get_config`. Returns: A `tf.keras.constraints.Constraint` instance. r r )clsconfigs r from_configConstraint.from_configUs$}V}rr N) __name__ __module__ __qualname____firstlineno____doc__rr classmethodr__static_attributes__r rrr r !s%0  rr zkeras.constraints.MaxNormzkeras.constraints.max_normcj\rSrSrSrSSjr\RS5r\RS5r Sr g) MaxNormja MaxNorm weight constraint. Constrains the weights incident to each hidden unit to have a norm less than or equal to a desired value. Also available via the shortcut function `tf.keras.constraints.max_norm`. Args: max_value: the maximum norm value for the incoming weights. axis: integer, axis along which to calculate weight norms. For instance, in a `Dense` layer the weight matrix has shape `(input_dim, output_dim)`, set `axis` to `0` to constrain each weight vector of length `(input_dim,)`. In a `Conv2D` layer with `data_format="channels_last"`, the weight tensor has shape `(rows, cols, input_depth, output_depth)`, set `axis` to `[0, 1, 2]` to constrain the weights of each filter tensor of size `(rows, cols, input_depth)`. cXlX lgN max_valueaxis)rr+r,s r__init__MaxNorm.__init__s " rc[R"[R"[R"U5UR SS95n[R "USUR5nX[R"5U-- -$)NTr,keepdimsr) rsqrttf reduce_sumsquarer,clipr+epsilonrrnormsdesireds rrMaxNorm.__call__s^ MM"))A,TYY F ,,ua8w0589::rc4URURS.$)Nr*r*rs rrMaxNorm.get_configs!^^TYY??r)r,r+N)r rrr r!r"r-rdo_not_generate_docsrrr$r rrr&r&jsB.&&;';&&@'@rr&zkeras.constraints.NonNegzkeras.constraints.non_negc\rSrSrSrSrSrg)NonNegzuConstrains the weights to be non-negative. Also available via the shortcut function `tf.keras.constraints.non_neg`. cU[R"[R"US5[R"55-$)N)r3cast greater_equalrfloatxrs rrNonNeg.__call__s,2772++As3W^^5EFFFrr N)rrr r!r"rr$r rrrBrBs  GrrBzkeras.constraints.UnitNormzkeras.constraints.unit_normcj\rSrSrSrSSjr\RS5r\RS5r Sr g) UnitNormaConstrains the weights incident to each hidden unit to have unit norm. Also available via the shortcut function `tf.keras.constraints.unit_norm`. Args: axis: integer, axis along which to calculate weight norms. For instance, in a `Dense` layer the weight matrix has shape `(input_dim, output_dim)`, set `axis` to `0` to constrain each weight vector of length `(input_dim,)`. In a `Conv2D` layer with `data_format="channels_last"`, the weight tensor has shape `(rows, cols, input_depth, output_depth)`, set `axis` to `[0, 1, 2]` to constrain the weights of each filter tensor of size `(rows, cols, input_depth)`. cXlgr)r,)rr,s rr-UnitNorm.__init__s rc U[R"5[R"[R"[R "U5UR SS95-- $)NTr0)rr7r2r3r4r5r,rs rrUnitNorm.__call__sF OO ll biilTJ   rcSUR0$)Nr,rNrs rrUnitNorm.get_configs ""rrNN)rr?r rrrKrKs@$&& ' &&#'#rrKzkeras.constraints.MinMaxNormzkeras.constraints.min_max_normcj\rSrSrSrSSjr\RS5r\RS5r Sr g) MinMaxNormaMinMaxNorm weight constraint. Constrains the weights incident to each hidden unit to have the norm between a lower bound and an upper bound. Also available via the shortcut function `tf.keras.constraints.min_max_norm`. Args: min_value: the minimum norm for the incoming weights. max_value: the maximum norm for the incoming weights. rate: rate for enforcing the constraint: weights will be rescaled to yield `(1 - rate) * norm + rate * norm.clip(min_value, max_value)`. Effectively, this means that rate=1.0 stands for strict enforcement of the constraint, while rate<1.0 means that weights will be rescaled at each step to slowly move towards a value inside the desired interval. axis: integer, axis along which to calculate weight norms. For instance, in a `Dense` layer the weight matrix has shape `(input_dim, output_dim)`, set `axis` to `0` to constrain each weight vector of length `(input_dim,)`. In a `Conv2D` layer with `data_format="channels_last"`, the weight tensor has shape `(rows, cols, input_depth, output_depth)`, set `axis` to `[0, 1, 2]` to constrain the weights of each filter tensor of size `(rows, cols, input_depth)`. c4XlX lX0lX@lgr) min_valuer+rater,)rrYr+rZr,s rr-MinMaxNorm.__init__s""  rcb[R"[R"[R"U5UR SS95nUR [R"X RUR5-SUR - U--nX[R"5U-- -$)NTr0) rr2r3r4r5r,rZr6rYr+r7r8s rrMinMaxNorm.__call__s MM"))A,TYY F  II UNNDNNK K499}% & w0589::rc`URURURURS.$)NrXrXrs rrMinMaxNorm.get_configs*IIII   r)r,r+rYrZN)rE?rarr?r rrrUrUs@> &&;';&& ' rrUz"keras.constraints.RadialConstraintz#keras.constraints.radial_constraintcB\rSrSrSr\R S5rSrSr g)RadialConstraintaIConstrains `Conv2D` kernel weights to be the same for each radius. Also available via the shortcut function `tf.keras.constraints.radial_constraint`. For example, the desired output for the following 4-by-4 kernel: ``` kernel = [[v_00, v_01, v_02, v_03], [v_10, v_11, v_12, v_13], [v_20, v_21, v_22, v_23], [v_30, v_31, v_32, v_33]] ``` is this:: ``` kernel = [[v_11, v_11, v_11, v_11], [v_11, v_33, v_33, v_11], [v_11, v_33, v_33, v_11], [v_11, v_11, v_11, v_11]] ``` This constraint can be applied to any `Conv2D` layer version, including `Conv2DTranspose` and `SeparableConv2D`, and with either `"channels_last"` or `"channels_first"` data format. The method assumes the weight tensor is of shape `(rows, cols, input_depth, output_depth)`. c URnURbURS:wa[SU35eUup4pV[R"XXEU-45n[R "UR [R"[R"USS9SS95n[R"[R"[R"USS9SS9X4XV45$)NzGThe weight tensor must have rank 4. Received weight tensor with shape: rNr) shaperank ValueErrorrreshapemap_fn_kernel_constraintstackr3unstack)rrw_shapeheightwidthchannelskernelss rrRadialConstraint.__call__!s'' << 7<<1#466=Y@  ,3(x OOA'/AB C NN  # # MM"**QR0q 9  MM"**QQ/b 9 H .  rc ^^^ [R"SS/SS//SS9m[R"T5Sn[R"US- S5m [R"[R"[ R RUS5S5UU 4SjUU 4Sj5n[R"[R"[ R RUS5S5S S 5nU 4S jnUUU 4S jn[ RRRUUXC/UR5[ R"S S /5/S9upsU$)zCRadially constraints a kernel with shape (height, width, channels).r]int32dtyperr>boolc&>TTS- T2TS- T24$)Nr]r kernelstartsr5RadialConstraint._kernel_constraint..As!F519u,eai%.??@rcf>TTS- T2TS- T24[R"STRS9-$)Nr])r>r>rx)rzerosryr|srrrBs8F519u,eai%.??@mmF&,,78rc,[R"SSS9$)Nrrwrxrconstantr rrrrGG$$Qg6rc,[R"SSS9$)Nr]rwrxrr rrrrHrrc2>[R"UT5$r))rless)indexargsr~s rrrJsw||E5/Irc P>US-[R"UTTTU-TU-4S94$)Nr])constant_values)r3pad)iarrayr}paddingr~s rbody_fn4RadialConstraint._kernel_constraint..body_fnLs8q5"&&wuqy%!)7K0L rN)shape_invariants) rrrhrFswitchr3mathfloormodcompatv1 while_loop get_shape TensorShape) rr} kernel_shape kernel_newrwhile_conditionr_rr~s ` @@rrm#RadialConstraint._kernel_constraint7s""QFQF#37C}}V,Q/  \A-w7^^ LL)),:F C @ 8  LL)),:F C 6 6  J   //    #oo/t 1MN 0  rr N) rrr r!r"rr@rrmr$r rrrcrcs&:&& ' * rrczkeras.constraints.serializecUcg[U[5(d#[R"S[ U5S35 U(a[ R "U5$[ U5$)NzThe `keras.constraints.serialize()` API should only be used for objects of type `keras.constraints.Constraint`. Found an instance of type z+, which may lead to improper serialization.) isinstancer warningswarntypelegacy_serializationr) constraintuse_legacy_formats r serializerhs_ j* - -  J'()  #:::FF !* --rzkeras.constraints.deserializecvU(a[R"U[5USS9$[U[5USS9$)Nr)module_objectscustom_objectsprintable_module_name)rrglobals)rrrs r deserializerxsC#<< "9)".   $y%*  rzkeras.constraints.getcUcg[U[5(aSU;n[XS9$[U[5(a[U50S.n[ U5$[ U5(aU$[ SU35e)z)Retrieves a TF-Keras constraint function.Nmodule)r) class_namerz4Could not interpret constraint function identifier: )rdictrstrgetcallablerj) identifierrrs rrrs}*d##$J6:KK J $ $ #J2>6{ *  B:, O  r)F)NF)#r"rtensorflow.compat.v2rv2r3 tf_keras.srcrtf_keras.src.saving.legacyrr%tf_keras.src.saving.serialization_librr tensorflow.python.util.tf_exportrtensorflow.tools.docsrr r&rBrKrUrcmax_normnon_neg unit_norm min_max_normradial_constraintmaxnormnonnegunitnormrrrr rrrs"G!! LJH:.,-EE.EP)+GH&@j&@I&@R(*EFGZGGG*,IJ!#z!#K!#H,.NO8 8 P8 v(*OTzTTr     $   +, .- .-. /  %& ' r