o 2h@s~ddlZddlZddlmZddlmZddlm Z ddl m Z ddl m Z edGdd d Zd d Zd d ZddZdS)N)backend) keras_export) global_state) jax_utils) auto_namezkeras.random.SeedGeneratorc@s8eZdZdZd ddZd ddZdd Zed d ZdS) SeedGeneratoraRGenerates variable seeds upon each call to a function generating random numbers. In Keras, all random number generators (such as `keras.random.normal()`) are stateless, meaning that if you pass an integer seed to them (such as `seed=42`), they will return the same values for repeated calls. To get different values for each call, a `SeedGenerator` providing the state of the random generator has to be used. Note that all the random number generators have a default seed of None, which implies that an internal global SeedGenerator is used. If you need to decouple the RNG from the global state you can provide a local `StateGenerator` with either a deterministic or random initial state. Remark concerning the JAX backen: Note that the use of a local `StateGenerator` as seed argument is required for JIT compilation of RNG with the JAX backend, because the use of global state is not supported. Example: ```python seed_gen = keras.random.SeedGenerator(seed=42) values = keras.random.normal(shape=(2, 3), seed=seed_gen) new_values = keras.random.normal(shape=(2, 3), seed=seed_gen) ``` Usage in a layer: ```python class Dropout(keras.Layer): def __init__(self, **kwargs): super().__init__(**kwargs) self.seed_generator = keras.random.SeedGenerator(1337) def call(self, x, training=False): if training: return keras.random.dropout( x, rate=0.5, seed=self.seed_generator ) return x ``` Nc s|dur tjj}|_|dd}|rtd||dur$|_nt__dur1tt t s=tdfdd}jj jdjj |dj dd d d _WddS1siwYdS) Nrz Unrecognized keyword arguments: z3Argument `seed` must be an integer. Received: seed=cs |dd}jjdg|dS)Ndtyperr)getrconvert_to_tensor)argskwargsrseedselfZ/var/www/html/chatgem/venv/lib/python3.10/site-packages/keras/src/random/seed_generator.pyseed_initializerRs z0SeedGenerator.__init__..seed_initializer)caller)Fnoneseed_generator_state)shaper trainable aggregationname)r __class____name__rpop ValueErrorr _initial_seedmake_default_seed isinstanceint name_scopeVariablerandom_seed_dtypestate)rrrr custom_backendrrrr__init__<s6    "zSeedGenerator.__init__TcCsh|j}|jd}|r&|jjtddg|jd}|j|jj |||S|j|ddd|S)Nrr i iO=) r'valuerr nparrayrassignnumpyadd)rordered seed_statenew_seed_value incrementrrrnext`s zSeedGenerator.nextcCs d|jiS)Nr)r )rrrr get_configos zSeedGenerator.get_configcCs|di|S)Nrr)clsconfigrrr from_configrszSeedGenerator.from_config)NN)T) r __module__ __qualname____doc__r)r5r6 classmethodr9rrrrr s . $rcCs8trtdtd}|durt}td||S)Na[JAX RNG] When tracing a JAX function, you should only use seeded random ops, e.g. you should create a `SeedGenerator` instance, attach it to your layer/model, and pass the instance as the `seed` argument when calling random ops. Unseeded random ops would get incorrectly traced by JAX and would become constant after tracing. Example: ``` # Make sure to set the seed generator as a layer attribute self.seed_generator = keras.random.SeedGenerator(seed=1337) ... out = keras.random.normal(shape=(1,), seed=self.seed_generator) ```global_seed_generator)ris_in_jax_tracing_scoperrget_global_attributerset_global_attribute)genrrrr>ws  r>cCstdtdS)Nr*geA) python_randomrandintr#rrrrr!sr!cCsvddlm}ddlm}t|tr|St|tr#||dg|dS|dur.tjddStd|dt |d ) Nr)r )r&r F)r1z\Argument `seed` must be either an integer or an instance of `SeedGenerator`. Received: seed=z (of type )) keras.src.backendr r&r"rr5r#r>rtype)rr r&rrr draw_seeds    rH)randomrCr/r, keras.srcrkeras.src.api_exportrkeras.src.backend.commonrkeras.src.utilsrkeras.src.utils.namingrrr>r!rHrrrrs     j