Aprenda o que há de mais recente em aprendizado de máquina, IA generativa e muito mais no WiML Symposium 2023
Registre-se
Mantenha tudo organizado com as coleções
Salve e categorize o conteúdo com base nas suas preferências.
tensorflow :: ops :: ResourceApplyRMSProp
#include <training_ops.h>
Atualize '* var' de acordo com o algoritmo RMSProp.
Resumo
Observe que na implementação densa deste algoritmo, ms e mom serão atualizados mesmo se o grad for zero, mas nesta implementação esparsa, ms e mom não serão atualizados em iterações durante as quais o grad for zero.
quadrado_médio = decadência * quadrado_médio + (1-decaimento) * gradiente ** 2 Delta = taxa_de_prendizagem * gradiente / quadrado (quadrado_médio + épsilon)
ms <- rho * ms_ {t-1} + (1-rho) * grad * grad mom <- momentum * mom_ {t-1} + lr * grad / sqrt (ms + epsilon) var <- var - mom
Argumentos:
- escopo: um objeto Scope
- var: deve ser de uma variável ().
- ms: deve ser de uma variável ().
- mãe: deve ser de uma variável ().
- lr: Fator de escala. Deve ser um escalar.
- rho: Taxa de decaimento. Deve ser um escalar.
- epsilon: termo de cume. Deve ser um escalar.
- grad: O gradiente.
Atributos opcionais (consulte Attrs
):
- use_locking: se
True
, a atualização dos tensores var, ms e mom é protegida por um bloqueio; caso contrário, o comportamento é indefinido, mas pode exibir menos contenção.
Retorna:
Construtores e Destruidores |
---|
ResourceApplyRMSProp (const :: tensorflow::Scope & scope, :: tensorflow::Input var, :: tensorflow::Input ms, :: tensorflow::Input mom, :: tensorflow::Input lr, :: tensorflow::Input rho, :: tensorflow::Input momentum, :: tensorflow::Input epsilon, :: tensorflow::Input grad)
|
ResourceApplyRMSProp (const :: tensorflow::Scope & scope, :: tensorflow::Input var, :: tensorflow::Input ms, :: tensorflow::Input mom, :: tensorflow::Input lr, :: tensorflow::Input rho, :: tensorflow::Input momentum, :: tensorflow::Input epsilon, :: tensorflow::Input grad, const ResourceApplyRMSProp::Attrs & attrs) |
Atributos públicos
Funções públicas
operador :: tensorflow :: Operação
operator::tensorflow::Operation() const
Funções estáticas públicas
UseLocking
Attrs UseLocking(
bool x
)
Exceto em caso de indicação contrária, o conteúdo desta página é licenciado de acordo com a Licença de atribuição 4.0 do Creative Commons, e as amostras de código são licenciadas de acordo com a Licença Apache 2.0. Para mais detalhes, consulte as políticas do site do Google Developers. Java é uma marca registrada da Oracle e/ou afiliadas.
Última atualização 2020-04-20 UTC.
[null,null,["Última atualização 2020-04-20 UTC."],[],[],null,["# tensorflow::ops::ResourceApplyRMSProp Class Reference\n\ntensorflow::ops::ResourceApplyRMSProp\n=====================================\n\n`#include \u003ctraining_ops.h\u003e`\n\nUpdate '\\*var' according to the RMSProp algorithm.\n\nSummary\n-------\n\nNote that in dense implementation of this algorithm, ms and mom will update even if the grad is zero, but in this sparse implementation, ms and mom will not update in iterations during which the grad is zero.\n\nmean_square = decay \\* mean_square + (1-decay) \\* gradient \\*\\* 2 Delta = learning_rate \\* gradient / sqrt(mean_square + epsilon)\n\nms \\\u003c- rho \\* ms_{t-1} + (1-rho) \\* grad \\* grad mom \\\u003c- momentum \\* mom_{t-1} + lr \\* grad / sqrt(ms + epsilon) var \\\u003c- var - mom\n\nArguments:\n\n- scope: A [Scope](/versions/r1.15/api_docs/cc/class/tensorflow/scope#classtensorflow_1_1_scope) object\n- var: Should be from a Variable().\n- ms: Should be from a Variable().\n- mom: Should be from a Variable().\n- lr: Scaling factor. Must be a scalar.\n- rho: Decay rate. Must be a scalar.\n- epsilon: Ridge term. Must be a scalar.\n- grad: The gradient.\n\n\u003cbr /\u003e\n\nOptional attributes (see [Attrs](/versions/r1.15/api_docs/cc/struct/tensorflow/ops/resource-apply-r-m-s-prop/attrs#structtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1_1_attrs)):\n\n- use_locking: If `True`, updating of the var, ms, and mom tensors is protected by a lock; otherwise the behavior is undefined, but may exhibit less contention.\n\n\u003cbr /\u003e\n\nReturns:\n\n- the created [Operation](/versions/r1.15/api_docs/cc/class/tensorflow/operation#classtensorflow_1_1_operation)\n\n\u003cbr /\u003e\n\n| ### Constructors and Destructors ||\n|---|---|\n| [ResourceApplyRMSProp](#classtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1ae91eb1e2b6b3e0c166963715954c5122)`(const ::`[tensorflow::Scope](/versions/r1.15/api_docs/cc/class/tensorflow/scope#classtensorflow_1_1_scope)` & scope, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` var, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` ms, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` mom, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` lr, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` rho, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` momentum, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` epsilon, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` grad)` ||\n| [ResourceApplyRMSProp](#classtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1a75df67ab1eea661cf727de50a0a7fb98)`(const ::`[tensorflow::Scope](/versions/r1.15/api_docs/cc/class/tensorflow/scope#classtensorflow_1_1_scope)` & scope, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` var, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` ms, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` mom, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` lr, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` rho, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` momentum, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` epsilon, ::`[tensorflow::Input](/versions/r1.15/api_docs/cc/class/tensorflow/input#classtensorflow_1_1_input)` grad, const `[ResourceApplyRMSProp::Attrs](/versions/r1.15/api_docs/cc/struct/tensorflow/ops/resource-apply-r-m-s-prop/attrs#structtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1_1_attrs)` & attrs)` ||\n\n| ### Public attributes ||\n|-------------------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------|\n| [operation](#classtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1a48b8adc2f5de282222027a49c23ff42d) | [Operation](/versions/r1.15/api_docs/cc/class/tensorflow/operation#classtensorflow_1_1_operation) |\n\n| ### Public functions ||\n|----------------------------------------------------------------------------------------------------------------------------------------|---------|\n| [operator::tensorflow::Operation](#classtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1afe6e89eae46d27e22c2ac94cc2c7aadc)`() const ` | ` ` ` ` |\n\n| ### Public static functions ||\n|------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| [UseLocking](#classtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1aacf915d8791a673d2e19b0af3d86af3a)`(bool x)` | [Attrs](/versions/r1.15/api_docs/cc/struct/tensorflow/ops/resource-apply-r-m-s-prop/attrs#structtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop_1_1_attrs) |\n\n| ### Structs ||\n|-----------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| [tensorflow::ops::ResourceApplyRMSProp::Attrs](/versions/r1.15/api_docs/cc/struct/tensorflow/ops/resource-apply-r-m-s-prop/attrs) | Optional attribute setters for [ResourceApplyRMSProp](/versions/r1.15/api_docs/cc/class/tensorflow/ops/resource-apply-r-m-s-prop#classtensorflow_1_1ops_1_1_resource_apply_r_m_s_prop). |\n\nPublic attributes\n-----------------\n\n### operation\n\n```text\nOperation operation\n``` \n\nPublic functions\n----------------\n\n### ResourceApplyRMSProp\n\n```gdscript\n ResourceApplyRMSProp(\n const ::tensorflow::Scope & scope,\n ::tensorflow::Input var,\n ::tensorflow::Input ms,\n ::tensorflow::Input mom,\n ::tensorflow::Input lr,\n ::tensorflow::Input rho,\n ::tensorflow::Input momentum,\n ::tensorflow::Input epsilon,\n ::tensorflow::Input grad\n)\n``` \n\n### ResourceApplyRMSProp\n\n```gdscript\n ResourceApplyRMSProp(\n const ::tensorflow::Scope & scope,\n ::tensorflow::Input var,\n ::tensorflow::Input ms,\n ::tensorflow::Input mom,\n ::tensorflow::Input lr,\n ::tensorflow::Input rho,\n ::tensorflow::Input momentum,\n ::tensorflow::Input epsilon,\n ::tensorflow::Input grad,\n const ResourceApplyRMSProp::Attrs & attrs\n)\n``` \n\n### operator::tensorflow::Operation\n\n```gdscript\n operator::tensorflow::Operation() const \n``` \n\nPublic static functions\n-----------------------\n\n### UseLocking\n\n```text\nAttrs UseLocking(\n bool x\n)\n```"]]