| /external/tensorflow/tensorflow/compiler/mlir/tensorflow/transforms/ |
| D | decompose_resource_ops.td | 224 // alpha <- learning_rate * sqrt(1 - beta2^t) / (1 - beta1^t)
225 // m_t <- beta1 * m_{t-1} + (1 - beta1) * g_t
232 $beta1, $beta2, $epsilon, $grad, BoolAttr:$_,
244 (TF_MulOp $beta1, (CreateTFReadVariableOp $src_op, $grad, $m_resource)),
245 (TF_MulOp (TF_SubOp $one, $beta1), $grad)
269 // alpha <- learning_rate * sqrt(1 - beta2^t) / (1 - beta1^t)
270 // m_t <- beta1 * m_{t-1} + (1 - beta1) * g_t
272 // variable <- variable - (alpha * (m_t * beta1 + (1 - beta1) * g_t) /
278 $beta1, $beta2, $epsilon, $grad, BoolAttr:$_,
290 (TF_MulOp $beta1, (CreateTFReadVariableOp $src_op, $grad, $m_resource)),
[all …]
|
| /external/tensorflow/tensorflow/core/api_def/base_api/ |
| D | api_def_ApplyAdaMax.pbtxt | 35 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)
|
| D | api_def_ResourceApplyAdaMax.pbtxt | 35 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)
|
| D | api_def_ResourceApplyAddSign.pbtxt | 55 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
|
| D | api_def_ResourceApplyPowerSign.pbtxt | 55 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
|
| D | api_def_ApplyAddSign.pbtxt | 61 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
|
| D | api_def_ApplyPowerSign.pbtxt | 61 m_t <- beta1 * m_{t-1} + (1 - beta1) * g
|
| /external/tensorflow/tensorflow/python/keras/optimizer_v2/ |
| D | nadam_test.py | 41 def update_m_cache(m_cache, t, beta1=0.9): argument
42 mu_t = beta1 * (1 - 0.5 * 0.96**(0.004 * (t + 1)))
54 beta1=0.9, argument
58 mu_t = beta1 * (1 - 0.5 * 0.96**(0.004 * (t + 1)))
59 mu_t_1 = beta1 * (1 - 0.5 * 0.96**(0.004 * (t + 2)))
62 m_t = beta1 * m + (1 - beta1) * g_t
|
| D | adamax_test.py | 42 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)))
|
| D | adam_test.py | 44 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
220 beta1 = lambda: 0.9 function
[all …]
|
| /external/tensorflow/tensorflow/core/ops/compat/ops_history_v1/ |
| D | ApplyAdam.pbtxt | 31 name: "beta1"
111 name: "beta1"
198 name: "beta1"
287 name: "beta1"
377 name: "beta1"
|
| D | ResourceApplyAdam.pbtxt | 28 name: "beta1"
101 name: "beta1"
181 name: "beta1"
263 name: "beta1"
346 name: "beta1"
|
| D | ResourceApplyAdaMax.pbtxt | 24 name: "beta1"
|
| D | ResourceApplyAdamWithAmsgrad.pbtxt | 32 name: "beta1"
|
| /external/tensorflow/tensorflow/core/ops/compat/ops_history_v2/ |
| D | ApplyAdam.pbtxt | 31 name: "beta1"
111 name: "beta1"
198 name: "beta1"
287 name: "beta1"
377 name: "beta1"
|
| D | ResourceApplyAdam.pbtxt | 28 name: "beta1"
101 name: "beta1"
181 name: "beta1"
263 name: "beta1"
346 name: "beta1"
|
| D | ResourceApplyAdaMax.pbtxt | 24 name: "beta1"
|
| D | ApplyAdaMax.pbtxt | 27 name: "beta1"
|
| D | ResourceApplyAdamWithAmsgrad.pbtxt | 32 name: "beta1"
|
| /external/tensorflow/tensorflow/compiler/tests/ |
| D | adam_test.py | 40 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/ |
| D | adam.py | 43 beta1=0.9, argument
106 self._beta1 = beta1
143 beta1 = self._call_if_callable(self._beta1)
148 self._beta1_t = ops.convert_to_tensor(beta1, name="beta1")
|
| D | adam_test.py | 44 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
184 beta1 = lambda: 0.9 function
189 beta1 = beta1()
|
| D | training_ops_test.py | 425 beta1 = np.array(0.9, dtype=var.dtype)
427 beta1_power = beta1**t
431 beta1_t = constant_op.constant(beta1, self._toType(var.dtype), [])
440 new_var, _, _ = self._adamUpdateNumpy(var, grad, t, m, v, lr, beta1,
449 def _adamUpdateNumpy(self, param, g_t, t, m, v, alpha, beta1, beta2, epsilon): argument
450 alpha_t = alpha * np.sqrt(1 - beta2**t) / (1 - beta1**t)
452 m_t = beta1 * m + (1 - beta1) * g_t
|
| /external/tensorflow/tensorflow/python/tpu/ |
| D | tpu_embedding.py | 559 beta1: float = 0.9,
610 if beta1 < 0. or beta1 >= 1.:
611 raise ValueError('beta1 must be between 0. and 1; got {}.'.format(beta1))
620 self.beta1 = beta1
760 beta1: float = 0.9,
814 if beta1 < 0. or beta1 >= 1.:
815 raise ValueError('beta1 must be between 0. and 1; got {}.'.format(beta1))
827 self.beta1 = beta1
2174 table_descriptor.optimization_parameters.adam.beta1 = (
2175 self._optimization_parameters.beta1)
[all …]
|
| /external/tensorflow/tensorflow/tools/api/golden/v1/ |
| D | tensorflow.tpu.experimental.-adam-parameters.pbtxt | 8 …argspec: "args=[\'self\', \'learning_rate\', \'beta1\', \'beta2\', \'epsilon\', \'lazy_adam\', \'s…
|