HomeSort by relevance Sort by last modified time
    Searched refs:log_scale (Results 1 - 5 of 5) sorted by null

  /external/tensorflow/tensorflow/contrib/distributions/python/ops/bijectors/
real_nvp.py 138 `log_scale` from both the forward domain (`x`) and the inverse domain
144 `log_scale` is equivalent to (but more efficient than) returning zero.
146 implementation assumes `log_scale` does not depend on the forward domain
184 shift, log_scale = self._shift_and_log_scale_fn(
187 if log_scale is not None:
188 y1 *= math_ops.exp(log_scale)
198 shift, log_scale = self._shift_and_log_scale_fn(
203 if log_scale is not None:
204 x1 *= math_ops.exp(-log_scale)
211 _, log_scale = self._shift_and_log_scale_fn
    [all...]
masked_autoregressive.py 104 shift, log_scale = shift_and_log_scale_fn(y)
105 y = x * math_ops.exp(log_scale) + shift
113 shift, log_scale = shift_and_log_scale_fn(y)
114 return (y - shift) / math_ops.exp(log_scale)
118 forward pass each calculation of `shift` and `log_scale` is based on the `y`
121 the "last" `y` used to compute `shift`, `log_scale`. (Roughly speaking, this
191 `log_scale` from both the forward domain (`x`) and the inverse domain
197 `log_scale` is equivalent to (but more efficient than) returning zero.
199 implementation assumes `log_scale` does not depend on the forward domain
229 shift, log_scale = self._shift_and_log_scale_fn(y
    [all...]
  /toolchain/binutils/binutils-2.27/gprof/
hist.c 570 unsigned log_scale;
605 log_scale = 5; /* Milli-seconds is BSD-default. */
611 log_scale = 0;
637 for (log_scale = 0; log_scale < ARRAY_SIZE (SItab); log_scale ++)
639 double scaled_value = SItab[log_scale].scale * top_time;
650 print_header (SItab[log_scale].prefix);
661 print_line (time_sorted_syms[sym_index], SItab[log_scale].scale);
569 unsigned log_scale; local
  /external/tensorflow/tensorflow/contrib/distributions/python/kernel_tests/bijectors/
real_nvp_test.py 135 log_scale = constant_op.constant([0.5])
136 return shift, log_scale
  /external/tensorflow/tensorflow/python/ops/distributions/
special_math.py 333 log_scale = -0.5 * x_2 - math_ops.log(-x) - 0.5 * math.log(2. * math.pi)
334 return log_scale + math_ops.log(_log_ndtr_asymptotic_series(x, series_order))

Completed in 304 milliseconds