Home
last modified time | relevance | path

Searched refs:beta1 (Results 1 – 25 of 65) sorted by relevance

123

/external/tensorflow/tensorflow/core/api_def/base_api/
Dapi_def_ApplyAdaMax.pbtxt35 name: "beta1"
74 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
76 variable <- variable - learning_rate / (1 - beta1^t) * m_t / (v_t + epsilon)
Dapi_def_ResourceApplyAdaMax.pbtxt35 name: "beta1"
68 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
70 variable <- variable - learning_rate / (1 - beta1^t) * m_t / (v_t + epsilon)
Dapi_def_ResourceApplyPowerSign.pbtxt55 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
Dapi_def_ResourceApplyAddSign.pbtxt55 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
Dapi_def_ApplyAddSign.pbtxt61 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
Dapi_def_ApplyPowerSign.pbtxt61 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
/external/tensorflow/tensorflow/contrib/opt/python/training/
Dnadam_optimizer_test.py38 beta1=0.9, argument
41 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
43 m_t = beta1 * m + (1 - beta1) * g_t
46 m_bar = (1 - beta1) * g_t + beta1 * m_t
Dadamax_test.py43 beta1=0.9, argument
46 m_t = beta1 * m + (1 - beta1) * g_t
48 param_t = param - (alpha / (1 - beta1**t)) * (m_t / (v_t + epsilon))
59 beta1=0.9, argument
63 m_t_slice = beta1 * m[indices] + (1 - beta1) * g_t
65 param_t_slice = param[indices] - ((alpha / (1 - beta1**t)) *
Dggt.py43 beta1=0.9, argument
112 self._set_hyper("beta1", beta1)
218 beta1 = state.get_hyper("beta1", dtype=var_dtype)
222 update_moment1 = moment1.assign(beta1 * moment1 + (1. - beta1) * flat_grad)
Dweight_decay_optimizers_test.py37 def adamw_update_numpy(param, g_t, t, m, v, lr=0.001, beta1=0.9, argument
39 lr_t = lr * np.sqrt(1 - beta2**t) / (1 - beta1**t)
41 m_t = beta1 * m + (1 - beta1) * g_t
Dlazy_adam_optimizer_test.py44 beta1=0.9, argument
47 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
49 m_t = beta1 * m + (1 - beta1) * g_t
187 beta1 = lambda: 0.9 function
192 beta1 = beta1()
Dadam_gs_optimizer.py43 beta1=0.9, argument
107 self._beta1 = beta1
131 beta1 = self._call_if_callable(self._beta1)
136 self._beta1_t = ops.convert_to_tensor(beta1, name="beta1")
Dadam_gs_optimizer_test.py44 beta1=0.9, argument
47 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
49 m_t = beta1 * m + (1 - beta1) * g_t
192 beta1 = lambda: 0.9 function
197 beta1 = beta1()
Dlazy_adam_gs_optimizer_test.py44 beta1=0.9, argument
47 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
49 m_t = beta1 * m + (1 - beta1) * g_t
211 beta1 = lambda: 0.9 function
216 beta1 = beta1()
Dweight_decay_optimizers.py359 def __init__(self, weight_decay, learning_rate=0.001, beta1=0.9, beta2=0.999, argument
380 weight_decay, learning_rate=learning_rate, beta1=beta1, beta2=beta2,
/external/tensorflow/tensorflow/python/keras/optimizer_v2/
Dnadam_test.py43 def update_m_cache(m_cache, t, beta1=0.9): argument
44 mu_t = beta1 * (1 - 0.5 * 0.96**(0.004 * (t + 1)))
56 beta1=0.9, argument
60 mu_t = beta1 * (1 - 0.5 * 0.96**(0.004 * (t + 1)))
61 mu_t_1 = beta1 * (1 - 0.5 * 0.96**(0.004 * (t + 2)))
64 m_t = beta1 * m + (1 - beta1) * g_t
Dadam_test.py44 beta1=0.9, argument
47 lr_t = lr * np.sqrt(1 - beta2**(t + 1)) / (1 - beta1**(t + 1))
49 m_t = beta1 * m + (1 - beta1) * g_t
63 beta1=0.9, argument
66 lr_t = lr * np.sqrt(1 - beta2**(t + 1)) / (1 - beta1**(t + 1))
68 m_t = beta1 * m + (1 - beta1) * g_t
84 beta1=0.9, argument
89 lr_t = lr * np.sqrt(1 - beta2**(t + 1)) / (1 - beta1**(t + 1))
90 m_t_slice = beta1 * m[indices] + (1 - beta1) * g_t
221 beta1 = lambda: 0.9 function
[all …]
Dadamax_test.py42 beta1=0.9, argument
45 m_t = beta1 * m + (1 - beta1) * g_t
47 param_t = param - (alpha / (1 - beta1**(t + 1))) * (m_t / (v_t + epsilon))
58 beta1=0.9, argument
62 m_t_slice = beta1 * m[indices] + (1 - beta1) * g_t
65 (alpha / (1 - beta1**(t + 1))) * (m_t_slice / (v_t_slice + epsilon)))
/external/tensorflow/tensorflow/core/kernels/
Dtraining_ops_gpu.cu.cc132 typename TTypes<T>::ConstScalar beta1, in operator ()()
141 m + (beta1.constant(one) - beta1).reshape(single).broadcast(bcast) * in operator ()()
153 (m * beta1.reshape(single).broadcast(bcast) + in operator ()()
154 (beta1.constant(one) - beta1).reshape(single).broadcast(bcast) * in operator ()()
176 typename TTypes<T>::ConstScalar beta1, in operator ()()
185 m + (beta1.constant(one) - beta1).reshape(single).broadcast(bcast) * in operator ()()
207 typename TTypes<T>::ConstScalar beta1, in operator ()()
216 m + (beta1.constant(one) - beta1).reshape(single).broadcast(bcast) * in operator ()()
/external/tensorflow/tensorflow/compiler/tests/
Dadamax_test.py39 beta1=0.9, argument
42 m_t = beta1 * m + (1 - beta1) * g_t
44 param_t = param - (alpha / (1 - beta1**t)) * (m_t / (v_t + epsilon))
Dadam_test.py40 beta1=0.9, argument
43 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
45 m_t = beta1 * m + (1 - beta1) * g_t
/external/tensorflow/tensorflow/python/training/
Dtraining_ops_test.py276 beta1 = np.array(0.9, dtype=var.dtype)
278 beta1_power = beta1**t
282 beta1_t = constant_op.constant(beta1, self._toType(var.dtype), [])
291 new_var, _, _ = self._adamUpdateNumpy(var, grad, t, m, v, lr, beta1,
300 def _adamUpdateNumpy(self, param, g_t, t, m, v, alpha, beta1, beta2, epsilon): argument
301 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
303 m_t = beta1 * m + (1 - beta1) * g_t
Dadam_test.py43 beta1=0.9, argument
46 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
48 m_t = beta1 * m + (1 - beta1) * g_t
183 beta1 = lambda: 0.9 function
188 beta1 = beta1()
Dadam.py41 beta1=0.9, argument
99 self._beta1 = beta1
136 beta1 = self._call_if_callable(self._beta1)
141 self._beta1_t = ops.convert_to_tensor(beta1, name="beta1")
/external/tensorflow/tensorflow/python/tpu/
Dtpu_embedding.py153 beta1=0.9, argument
179 if beta1 < 0. or beta1 >= 1.:
180 raise ValueError('beta1 must be between 0. and 1; got {}.'.format(beta1))
189 self.beta1 = beta1
880 table_descriptor.optimization_parameters.adam.beta1 = (
881 self._optimization_parameters.beta1)

123