| /third_party/mindspore/mindspore-src/source/mindspore/ccsrc/plugin/device/gpu/kernel/cuda_impl/cuda_ops/ |
| D | adagrad_v2_impl.cu | 32 const S *learning_rate, const T *gradient) { in ApplyAdagradV2Kernel()
43 const half *learning_rate, const half *gradient) { in ApplyAdagradV2Kernel()
54 const float *learning_rate, const half *gradient) { in ApplyAdagradV2Kernel()
65 const double *learning_rate, const half *gradient) { in ApplyAdagradV2Kernel()
76 const half *learning_rate, const double *gradient) { in ApplyAdagradV2Kernel()
87 const float *learning_rate, const double *gradient) { in ApplyAdagradV2Kernel()
98 const half *learning_rate, const float *gradient) { in ApplyAdagradV2Kernel()
109 const double *learning_rate, const float *gradient) { in ApplyAdagradV2Kernel()
120 const S *learning_rate, const T *gradient) { in ApplyAdagradV2Kernel_()
128 const half *learning_rate, const half *gradient) { in ApplyAdagradV2Kernel_()
[all …]
|
| D | adagrad_impl.cu | 84 …lobal__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const S *learning_rate, in ApplyAdagradKernel()
95 …l__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const float *learning_rate, in ApplyAdagradKernel()
106 …al__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const half *learning_rate, in ApplyAdagradKernel()
117 …__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const double *learning_rate, in ApplyAdagradKernel()
128 …__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const double *learning_rate, in ApplyAdagradKernel()
139 …__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const double *learning_rate, in ApplyAdagradKernel()
150 …al__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const half *learning_rate, in ApplyAdagradKernel()
161 …l__ void ApplyAdagradKernel(const size_t size, const bool update_slots, const float *learning_rate, in ApplyAdagradKernel()
172 cudaError_t ApplyAdagrad(const size_t size, const bool update_slots, const S *learning_rate, const … in ApplyAdagrad()
|
| D | adadelta_impl.cu | 31 __global__ void ApplyAdadeltaKernal(const size_t size, const S *learning_rate, const S *rho, const … in ApplyAdadeltaKernal()
44 __global__ void ApplyAdadeltaKernal(const size_t size, const float *learning_rate, const float *rho, in ApplyAdadeltaKernal()
60 __global__ void ApplyAdadeltaKernal(const size_t size, const float *learning_rate, const float *rho, in ApplyAdadeltaKernal()
75 __global__ void ApplyAdadeltaKernal(const size_t size, const half *learning_rate, const half *rho, … in ApplyAdadeltaKernal()
91 __global__ void ApplyAdadeltaKernal(const size_t size, const float *learning_rate, const float *rho, in ApplyAdadeltaKernal()
108 __global__ void ApplyAdadeltaKernal(const size_t size, const half *learning_rate, const half *rho, … in ApplyAdadeltaKernal()
123 __global__ void ApplyAdadeltaKernal(const size_t size, const half *learning_rate, const half *rho, … in ApplyAdadeltaKernal()
140 __global__ void ApplyAdadeltaKernal(const size_t size, const half *learning_rate, const half *rho, … in ApplyAdadeltaKernal()
157 __global__ void ApplyAdadeltaKernal(const size_t size, const float *learning_rate, const float *rho, in ApplyAdadeltaKernal()
171 __global__ void ApplyAdadeltaKernal(const size_t size, const float *learning_rate, const float *rho, in ApplyAdadeltaKernal()
[all …]
|
| D | apply_add_sign_impl.cu | 26 …__ void ApplyAddSignKernel(const size_t size, T *variable, T *accumulation, const S *learning_rate, in ApplyAddSignKernel()
36 …plyAddSignKernel(const size_t size, half *variable, half *accumulation, const float *learning_rate, in ApplyAddSignKernel()
51 …yAddSignKernel(const size_t size, float *variable, float *accumulation, const float *learning_rate, in ApplyAddSignKernel()
63 …lyAddSignKernel(const size_t size, float *variable, float *accumulation, const half *learning_rate, in ApplyAddSignKernel()
75 …lyAddSignKernel(const size_t size, float *variable, float *accumulation, const half *learning_rate, in ApplyAddSignKernel()
88 …pplyAddSignKernel(const size_t size, half *variable, half *accumulation, const half *learning_rate, in ApplyAddSignKernel()
101 cudaError_t ApplyAddSign(const size_t size, T *variable, T *accumulation, const S *learning_rate, c… in ApplyAddSign()
|
| D | apply_power_sign_impl.cu | 42 …_ void ApplyPowerSignKernel(const size_t size, T *variable, T *accumulation, const S learning_rate, in ApplyPowerSignKernel()
52 …lyPowerSignKernel(const size_t size, half *variable, half *accumulation, const float learning_rate, in ApplyPowerSignKernel()
67 …PowerSignKernel(const size_t size, float *variable, float *accumulation, const float learning_rate, in ApplyPowerSignKernel()
79 …plyPowerSignKernel(const size_t size, half *variable, half *accumulation, const half learning_rate, in ApplyPowerSignKernel()
92 …yPowerSignKernel(const size_t size, float *variable, float *accumulation, const half learning_rate, in ApplyPowerSignKernel()
105 …yPowerSignKernel(const size_t size, float *variable, float *accumulation, const half learning_rate, in ApplyPowerSignKernel()
118 cudaError_t ApplyPowerSign(const size_t size, T *variable, T *accumulation, const S learning_rate, … in ApplyPowerSign()
|
| D | adamax_impl.cu | 42 __global__ void ApplyAdamaxKernal(const size_t size, const S *b1_power, const S *learning_rate, con… in ApplyAdamaxKernal()
53 …id ApplyAdamaxKernal(const size_t size, const float *b1_power, const float *learning_rate, const f… in ApplyAdamaxKernal()
66 …id ApplyAdamaxKernal(const size_t size, const float *b1_power, const float *learning_rate, const f… in ApplyAdamaxKernal()
78 …void ApplyAdamaxKernal(const size_t size, const half *b1_power, const half *learning_rate, const h… in ApplyAdamaxKernal()
91 …void ApplyAdamaxKernal(const size_t size, const half *b1_power, const half *learning_rate, const h… in ApplyAdamaxKernal()
104 …id ApplyAdamaxKernal(const size_t size, const float *b1_power, const float *learning_rate, const f… in ApplyAdamaxKernal()
116 …id ApplyAdamaxKernal(const size_t size, const float *b1_power, const float *learning_rate, const f… in ApplyAdamaxKernal()
129 …id ApplyAdamaxKernal(const size_t size, const float *b1_power, const float *learning_rate, const f… in ApplyAdamaxKernal()
141 …void ApplyAdamaxKernal(const size_t size, const half *b1_power, const half *learning_rate, const h… in ApplyAdamaxKernal()
154 …void ApplyAdamaxKernal(const size_t size, const half *b1_power, const half *learning_rate, const h… in ApplyAdamaxKernal()
[all …]
|
| D | rmsprop_impl.cu | 71 …__ void RmsPropKernel(const size_t batch_size, const size_t input_elements, const T *learning_rate, in RmsPropKernel()
87 const Complex<double> *learning_rate, const Complex<double> *decay, in RmsPropKernel()
102 …opKernel(const size_t batch_size, const size_t input_elements, const Complex<float> *learning_rate, in RmsPropKernel()
117 cudaError_t RmsProp(const size_t batch_size, const size_t input_elements, const T *learning_rate, c… in RmsProp()
127 …d RmsPropCenterKernel(const size_t batch_size, const size_t input_elements, const T *learning_rate, in RmsPropCenterKernel()
145 …sPropCenterKernel(const size_t batch_size, const size_t input_elements, const float *learning_rate, in RmsPropCenterKernel()
163 … const Complex<double> *learning_rate, const Complex<double> *decay, in RmsPropCenterKernel()
184 … const Complex<float> *learning_rate, const Complex<float> *decay, in RmsPropCenterKernel()
204 …msPropCenterKernel(const size_t batch_size, const size_t input_elements, const half *learning_rate, in RmsPropCenterKernel()
224 …PropCenter(const size_t batch_size, const size_t input_elements, const T *learning_rate, const T *… in RmsPropCenter()
|
| D | adam_impl.cu | 32 … const T *beta2_power, const T *learning_rate, const T *beta1, const T *beta2, in ApplyAdamKernel()
51 …__ void AdamWeightDecayKernel(const size_t size, const float *gradient, const float *learning_rate, in AdamWeightDecayKernel()
65 …l__ void AdamWeightDecayKernel(const size_t size, const half *gradient, const float *learning_rate, in AdamWeightDecayKernel()
79 …l__ void AdamWeightDecayKernel(const size_t size, const half *gradient, const float *learning_rate, in AdamWeightDecayKernel()
98 … const T *beta2_power, const T *learning_rate, const T *beta1, const T *beta2, const T *epsilon, in ApplyAdam()
106 cudaError_t AdamWeightDecayOp(const size_t size, const S *gradient, const float *learning_rate, con… in AdamWeightDecayOp()
|
| D | momentum_impl.cu | 197 …r_t MomentumUpdateVariable(const size_t size, T *variable, T *accumulation, const S *learning_rate, in MomentumUpdateVariable()
211 … const S *learning_rate, const G *gradient, const S *momentum, in FusedWeightDecayScaleMomentum()
219 …sedScaleMomentum(const size_t size, S *scale, T *variable, T *accumulation, const S *learning_rate, in FusedScaleMomentum()
228 const S *learning_rate, const G *gradient, const S *momentum, in FusedWeightDecayMomentum()
238 … T **accumulation, S **learning_rate, G **gradient, S **momentum) { in CombineFusedMomentumScaleKernel()
249 … T **variable, T **accumulation, S **learning_rate, G **gradient, S **momentum, in CombineFusedScaleMomentum()
263 … S **learning_rate, G **gradient, S **momentum) { in CombineFusedMomentumWeightDecayScaleKernel()
276 S **learning_rate, G **gradient, S **momentum, in CombineFusedWeightDecayScaleMomentum()
|
| D | sparse_apply_centered_rms_prop_impl.cu | 37 … T *learning_rate, T *decay_rate, T *epsilon, T *momentum, in SparseApplyCenteredRMSPropUpdate()
59 … double *learning_rate, double *decay_rate, double *epsilon, in SparseApplyCenteredRMSPropUpdate()
82 … half *learning_rate, half *decay_rate, half *epsilon, half *momentum, in SparseApplyCenteredRMSPropUpdate()
112 … T *learning_rate, T *decay_rate, T *epsilon, T *momentum, const T *gradient, in CalSparseApplyCenteredRMSProp()
|
| D | sparse_apply_adagrad_impl.cu | 31 …d SparseApplyAdagradUpdate(const size_t size, const size_t indices_size, const float learning_rate, in SparseApplyAdagradUpdate()
51 …or_t CalSparseApplyAdagrad(const size_t size, const size_t indices_size, const float learning_rate, in CalSparseApplyAdagrad()
|
| /third_party/mindspore/mindspore-src/source/mindspore/python/mindspore/nn/ |
| D | dynamic_lr.py | 87 def _check_inputs(learning_rate, decay_rate, total_step, step_per_epoch, decay_epoch, is_stair): argument
98 def exponential_decay_lr(learning_rate, decay_rate, total_step, step_per_epoch, decay_epoch, is_sta… argument
153 def natural_exp_decay_lr(learning_rate, decay_rate, total_step, step_per_epoch, decay_epoch, is_sta… argument
209 def inverse_decay_lr(learning_rate, decay_rate, total_step, step_per_epoch, decay_epoch, is_stair=F… argument
328 def polynomial_decay_lr(learning_rate, end_learning_rate, total_step, step_per_epoch, decay_epoch, … argument
417 def warmup_lr(learning_rate, total_step, step_per_epoch, warmup_epoch): argument
|
| D | learning_rate_schedule.py | 51 def _check_inputs(learning_rate, decay_rate, decay_steps, is_stair, cls_name): argument
114 def __init__(self, learning_rate, decay_rate, decay_steps, is_stair=False): argument
185 def __init__(self, learning_rate, decay_rate, decay_steps, is_stair=False): argument
257 def __init__(self, learning_rate, decay_rate, decay_steps, is_stair=False): argument
404 …def __init__(self, learning_rate, end_learning_rate, decay_steps, power, update_decay_steps=False): argument
488 def __init__(self, learning_rate, warmup_steps): argument
|
| /third_party/mindspore/mindspore-src/source/mindspore/python/mindspore/nn/optim/ |
| D | ftrl.py | 31 def _tensor_run_opt_with_sparse_dist(opt, spars_opt, l1, l2, lr_power, learning_rate, linear, argument
45 def _apply_map_tensor_ftrl(l1, l2, lr_power, learning_rate, linear, weight, moment, indices, values… argument
83 def _run_map_tensor_opt_with_sparse_dist(opt, spars_opt, l1, l2, lr_power, learning_rate, linear, argument
102 def _tensor_run_opt_dist(opt, spars_opt, l1, l2, lr_power, learning_rate, linear, argument
116 def _tensor_run_opt_with_sparse(opt, spars_opt, l1, l2, lr_power, learning_rate, linear, argument
128 def _run_map_tensor_opt_with_sparse(opt, spars_opt, l1, l2, lr_power, learning_rate, linear, argument
139 def _tensor_run_opt(opt, spars_opt, l1, l2, lr_power, learning_rate, linear, argument
147 def _check_param(initial_accum, learning_rate, lr_power, l1, l2, use_locking, prim_name=None): argument
301 … def __init__(self, params, initial_accum=0.1, learning_rate=0.001, lr_power=-0.5, l1=0.0, l2=0.0, argument
368 def _init_distributed_opts(self, use_locking, learning_rate, l1, l2, lr_power): argument
|
| D | momentum.py | 33 def _tensor_run_opt_ext(opt, momentum, learning_rate, gradient, weight, moment): argument
40 def _tensor_run_opt_ext_dist(opt, momentum, learning_rate, gradient, weight, moment, argument
186 …def __init__(self, params, learning_rate, momentum, weight_decay=0.0, loss_scale=1.0, use_nesterov… argument
|
| D | rmsprop.py | 29 def _rmsprop_opt_(opt, decay, epsilon, momentum, learning_rate, weight, ms, mom, grad): argument
42 def _centered_rmsprop_opt_(opt, decay, epsilon, momentum, learning_rate, weight, mg, ms, mom, grad): argument
212 def __init__(self, params, learning_rate=0.1, decay=0.9, momentum=0.0, epsilon=1e-10, argument
|
| D | optimizer.py | 200 def __init__(self, learning_rate, parameters, weight_decay=0.0, loss_scale=1.0): argument
242 def _init_opt_attrs(self, learning_rate, parameters, weight_decay): argument
540 def _preprocess_single_lr(self, learning_rate): argument
569 def _build_single_lr(self, learning_rate, name): argument
609 def _parse_group_params(self, parameters, learning_rate): argument
640 def _init_group_params(self, parameters, learning_rate, weight_decay, grad_centralization): argument
789 def get_lr_value(learning_rate): argument
994 def __init__(self, learning_rate): argument
1008 def __init__(self, learning_rate, name): argument
|
| D | proximal_ada_grad.py | 30 def _tensor_run_opt_with_sparse(opt, sparse_opt, l1, l2, learning_rate, gradient, weight, accum): argument
38 def _tensor_run_opt(opt, sparse_opt, l1, l2, learning_rate, gradient, weight, accum): argument
191 def __init__(self, params, accum=0.1, learning_rate=0.001, l1=0.0, l2=0.0, argument
|
| D | ada_grad.py | 29 def _tensor_run_opt(opt, learning_rate, weight, accum, gradient): argument
194 def __init__(self, params, accum=0.1, learning_rate=0.001, argument
|
| D | sgd.py | 31 def _tensor_run_opt_ext(momentum, learning_rate, gradient, weight, accum, stat, opt): argument
160 …def __init__(self, params, learning_rate=0.1, momentum=0.0, dampening=0.0, weight_decay=0.0, neste… argument
|
| D | adadelta.py | 28 def _tensor_run_opt(opt, rho, epsilon, learning_rate, weight, accum, accum_update, gradient): argument
182 …def __init__(self, params, learning_rate=1.0, rho=0.9, epsilon=1e-6, loss_scale=1.0, weight_decay=… argument
|
| D | adamax.py | 35 def _tensor_run_opt(opt, beta1, beta2, beta1_power, eps, learning_rate, weight, moment1, moment2, g… argument
188 def __init__(self, params, learning_rate=0.001, beta1=0.9, beta2=0.999, eps=1e-08, argument
|
| /third_party/mindspore/mindspore-src/source/mindspore/lite/src/litert/kernel/cpu/fp32_grad/ |
| D | sgd.cc | 33 int DoSgd(float *weight, float *accumulate, float *gradient, float learning_rate, float dampening, … in DoSgd()
60 int DoSgdInit(float *weight, float *accumulate, float *gradient, float learning_rate, float moment,… in DoSgdInit()
91 float learning_rate = lr_; in DoExecute() local
115 float learning_rate = lr_; in ExecuteInit() local
231 float learning_rate = lr_; in OptimizerStep() local
|
| D | apply_momentum.cc | 32 static int DoApplyMomentum(float *weight, float *accumulate, float learning_rate, const float *grad… in DoApplyMomentum()
54 float learning_rate = lr_; in DoExecute() local
129 float learning_rate = lr_; in OptimizerStep() local
|
| /third_party/mindspore/mindspore-src/source/mindspore/ccsrc/plugin/device/cpu/kernel/ |
| D | rmsprop_cpu_kernel.cc | 48 … float *learning_rate, float *decay, float *momentum, in LaunchRMSPropUnuseCenter()
80 … T *mean_gradients, T *momentum, T *learning_rate, T *decay, in LaunchRMSPropUseCenter()
112 … std::complex<float> *learning_rate, std::complex<float> *decay, in LaunchRMSPropUseCenter()
145 … std::complex<double> *learning_rate, std::complex<double> *decay, in LaunchRMSPropUseCenter()
267 float *learning_rate = reinterpret_cast<float *>(inputs[kNumberThree]->device_ptr()); in LaunchKernel() local
283 T *learning_rate = static_cast<T *>(inputs[kNumberFive]->device_ptr()); in LaunchKernel() local
|