| /external/tensorflow/tensorflow/contrib/distributions/python/kernel_tests/ |
| geometric_test.py | 67 log_prob = geom.log_prob(x)
68 self.assertEqual([6,], log_prob.get_shape())
69 self.assertAllClose(expected_log_prob, log_prob.eval())
84 log_prob = geom.log_prob(x)
85 log_prob.eval(feed_dict=feed_dict)
88 log_prob = geom.log_prob(np.array([-1.], dtype=np.float32))
89 log_prob.eval(
[all...] |
| mvn_diag_plus_low_rank_test.py | 185 dist.log_prob(samps) - mvn_identity.log_prob(samps), 0)
189 dist.log_prob(samps) - mvn_scaled.log_prob(samps), 0)
193 dist.log_prob(samps) - mvn_diag.log_prob(samps), 0)
197 dist.log_prob(samps) - mvn_chol.log_prob(samps), 0)
208 baseline.log_prob(samps) - mvn_identity.log_prob(samps), 0
[all...] |
| vector_student_t_test.py | 50 def log_prob(self, x): member in class:_FakeVectorStudentT
69 return np.exp(self.log_prob(x))
93 self.assertAllClose(expected_mst.log_prob(x),
94 actual_mst.log_prob(x).eval(),
122 self.assertAllClose(expected_mst.log_prob(x),
123 actual_mst.log_prob(x).eval(),
155 self.assertAllClose(expected_mst.log_prob(x),
156 actual_mst.log_prob(x).eval(feed_dict=feed_dict),
186 self.assertAllClose(expected_mst.log_prob(x),
187 actual_mst.log_prob(x).eval()
[all...] |
| wishart_test.py | 213 self.assertAllClose(log_prob_df_seq, w.log_prob(chol_x).eval())
218 log_prob = np.array([
240 self.assertAllClose(log_prob[0], w.log_prob(x[0]).eval())
241 self.assertAllClose(log_prob[0:2], w.log_prob(x[0:2]).eval())
243 np.reshape(log_prob, (2, 2)),
244 w.log_prob(np.reshape(x, (2, 2, 2, 2))).eval())
246 np.reshape(np.exp(log_prob), (2, 2)),
249 w.log_prob(np.reshape(x, (2, 2, 2, 2))).get_shape()
[all...] |
| mixture_same_family_test.py | 44 log_prob_x = gm.log_prob(x)
55 log_prob_x = gm.log_prob(x)
67 log_prob_x = bm.log_prob(x)
81 log_prob_x = gm.log_prob(x)
96 log_prob_x = gm.log_prob(x)
|
| normal_conjugate_posteriors_test.py | 50 posterior_log_pdf = posterior.log_prob(x).eval()
71 posterior_log_pdf = posterior.log_prob(x).eval()
96 posterior_log_pdf = posterior.log_prob(x)
115 predictive_log_pdf = predictive.log_prob(x).eval()
|
| relaxed_bernoulli_test.py | 111 self.assertEqual(dist.probs.dtype, dist.log_prob([0.0]).dtype)
130 log_pdf = dist.log_prob(xs).eval()
137 self.assertAllClose(np.nan, dist.log_prob(0.0).eval())
138 self.assertAllClose([np.nan], [dist.log_prob(1.0).eval()])
|
| negative_binomial_test.py | 131 log_pmf = negbinom.log_prob(x)
150 log_pmf = negbinom.log_prob(x)
154 log_pmf = negbinom.log_prob([-1.])
159 log_pmf = negbinom.log_prob(x)
175 log_pmf = negbinom.log_prob(x)
251 log_prob_ = sess.run(nb.log_prob(x))
262 log_prob_ = sess.run(nb.log_prob(x))
|
| transformed_distribution_test.py | 72 for func in [[log_normal.log_prob, sp_dist.logpdf],
117 abs_normal.log_prob(2.13).eval())
150 sample_val, log_pdf_val = sess.run([sample, log_normal.log_prob(sample)])
178 sample_val, log_pdf_val = sess.run([sample, exp_normal.log_prob(sample)])
196 multi_logit_normal.log_prob(y).eval())
204 """Test log_prob when Jacobian and log_prob dtypes do not match."""
220 normal.log_prob(y).eval()
251 log_prob = exp2.log_prob(1.
[all...] |
| logistic_test.py | 51 log_prob = dist.log_prob(x)
52 self.assertEqual(log_prob.get_shape(), (6,))
53 self.assertAllClose(log_prob.eval(), expected_log_prob)
168 self.assertEqual(dist.loc.dtype, dist.log_prob(0.2).dtype)
|
| autoregressive_test.py | 89 td_log_prob_, ar_log_prob_ = sess.run([td.log_prob(x), ar.log_prob(x)])
|
| independent_test.py | 60 log_prob_x = ind.log_prob(x)
83 log_prob_x = ind.log_prob(x)
112 sample_entropy = -math_ops.reduce_mean(ind.log_prob(x), axis=0)
145 log_prob_x = ind.log_prob(x)
|
| sinh_arcsinh_test.py | 122 norm_lp, sasnorm_lp = sess.run([norm.log_prob(x), sasnorm.log_prob(x)])
164 norm_lp, sasnorm_lp = sess.run([norm.log_prob(x), sasnorm.log_prob(x)])
|
| poisson_test.py | 61 log_pmf = poisson.log_prob(x)
79 log_pmf = poisson.log_prob(x)
83 log_pmf = poisson.log_prob([-1.])
87 log_pmf = poisson.log_prob(x)
100 log_pmf = poisson.log_prob(x)
|
| mvn_tril_test.py | 55 log_pdf = mvn.log_prob(x)
75 log_pdf = mvn.log_prob(x)
96 log_pdf = mvn.log_prob(x)
180 # Check that log_prob(samples) works
181 log_prob_val = mvn.log_prob(samples_val).eval()
331 dist.log_prob(samps) - mvn_chol.log_prob(samps), 0)
|
| /external/tensorflow/tensorflow/python/kernel_tests/distributions/ |
| categorical_test.py | 101 dist.logits.dtype, dist.log_prob(np.array(
263 "cat_log_prob": cat.log_prob(disc_event_tf),
267 "norm_log_prob": norm.log_prob(real_event_tf),
288 self.assertAllClose(dist.log_prob([0, 1]).eval(), np.log([0.2, 0.4]))
289 self.assertAllClose(dist.log_prob([0.0, 1.0]).eval(), np.log([0.2, 0.4]))
397 log_prob = dist.log_prob([0, 1])
398 self.assertEqual(2, log_prob.get_shape().ndims)
399 self.assertAllEqual([1, 2], log_prob.get_shape())
401 log_prob = dist.log_prob([[[1, 1], [1, 0]], [[1, 0], [0, 1]]]
[all...] |
| bernoulli_test.py | 122 self.assertEqual(dist.probs.dtype, dist.log_prob(0).dtype)
151 self.assertAllClose(dist.log_prob(x).eval(), np.log(expected_pmf))
188 self.assertAllClose(np.log(0.5), dist.log_prob(1).eval({p: 0.5}))
190 np.log([0.5, 0.5, 0.5]), dist.log_prob([1, 1, 1]).eval({
194 np.log([0.5, 0.5, 0.5]), dist.log_prob(1).eval({
202 self.assertEqual(2, len(dist.log_prob(1).eval({p: [[0.5], [0.5]]}).shape))
206 self.assertEqual(2, len(dist.log_prob([[1], [1]]).eval().shape))
210 self.assertEqual((), dist.log_prob(1).get_shape())
211 self.assertEqual((1), dist.log_prob([1]).get_shape())
212 self.assertEqual((2, 1), dist.log_prob([[1], [1]]).get_shape()
[all...] |
| /external/tensorflow/tensorflow/contrib/bayesflow/python/kernel_tests/ |
| monte_carlo_test.py | 56 f=lambda x: x, log_p=p.log_prob, sampling_dist_q=q, n=n, seed=42)
60 f=math_ops.square, log_p=p.log_prob, sampling_dist_q=q, n=n, seed=42)
89 f=indicator, log_p=p.log_prob, sampling_dist_q=q, n=n, seed=42)
115 log_p=p.log_prob,
178 efx_reparam = mc.expectation(f, samples, p.log_prob)
179 efx_score = mc.expectation(f, samples, p.log_prob,
225 f=lambda x: p.log_prob(x) - q.log_prob(x),
227 log_prob=p.log_prob,
[all...] |
| csiszar_divergence_test.py | 494 p_log_prob=p.log_prob,
501 p_log_prob=p.log_prob,
528 p_log_prob=p.log_prob,
535 p_log_prob=p.log_prob,
563 p_log_prob=p.log_prob,
570 p_log_prob=p.log_prob,
601 p_log_prob=p.log_prob,
608 p_log_prob=p.log_prob,
643 p_log_prob=p.log_prob,
650 p_log_prob=p.log_prob,
[all...] |
| /external/tensorflow/tensorflow/contrib/bayesflow/python/ops/ |
| monte_carlo_impl.py | 68 For example, `f` works "just like" `q.log_prob`.
71 For example, `log_p` works "just like" `sampling_dist_q.log_prob`.
90 q_log_prob_z = q.log_prob(z)
145 For example, `log_f` works "just like" `sampling_dist_q.log_prob`.
148 For example, `log_p` works "just like" `q.log_prob`.
165 log_values = log_f(z) + log_p(z) - q.log_prob(z)
197 def expectation(f, samples, log_prob=None, use_reparametrization=True,
210 - `log(p(samples)) = log_prob(samples)` and
258 f=lambda x: p.log_prob(x) - q.log_prob(x)
[all...] |
| csiszar_divergence_impl.py | 855 takes `p_log_prob(q_samples) - q.log_prob(q_samples)`.
860 `reparameterization_type`, `sample(n, seed)`, and `log_prob(x)`.
901 f=lambda q_samples: f(p_log_prob(q_samples) - q.log_prob(q_samples)),
903 log_prob=q.log_prob, # Only used if use_reparametrization=False.
959 `log_prob(x)`. (In variational inference `q` is the approximate posterior
980 logqx = q.log_prob(x)
[all...] |
| /external/tensorflow/tensorflow/contrib/distributions/python/ops/ |
| conditional_transformed_distribution.py | 119 """Finish computation of log_prob on one element of the inverse image."""
121 log_prob = self.distribution.log_prob(x, **distribution_kwargs)
123 log_prob = math_ops.reduce_sum(log_prob, self._reduce_event_indices)
124 return math_ops.cast(ildj, log_prob.dtype) + log_prob
|
| conditional_distribution.py | 29 to their sample-based methods (i.e. `sample`, `log_prob`, etc.).
42 def log_prob(self, value, name="log_prob", **condition_kwargs): member in class:ConditionalDistribution
|
| /external/tensorflow/tensorflow/core/kernels/ |
| ctc_decoder_ops.cc | 58 Tensor** log_prob, OpOutputList* decoded_indices,
98 "log_probability", TensorShape({batch_size, top_paths_}), log_prob);
181 Tensor* log_prob = nullptr; variable
186 ctx, &inputs, &seq_len, &log_prob, &decoded_indices,
207 auto log_prob_t = log_prob->matrix<float>();
260 Tensor* log_prob = nullptr; variable
265 ctx, &inputs, &seq_len, &log_prob, &decoded_indices,
270 auto log_prob_t = log_prob->matrix<float>();
|
| /external/tensorflow/tensorflow/python/ops/distributions/ |
| transformed_distribution.py | 154 * `log_prob`
157 Programmatically: `(distribution.log_prob(bijector.inverse(y))
228 # mvn1.log_prob(x) == mvn2.log_prob(x)
432 """Finish computation of log_prob on one element of the inverse image."""
434 log_prob = self.distribution.log_prob(x)
436 log_prob = math_ops.reduce_sum(log_prob, self._reduce_event_indices)
437 log_prob += math_ops.cast(ildj, log_prob.dtype
[all...] |
