none;\n}\n\n.modal.fade .modal-dialog {\n  transition: -webkit-transform 0.3s ease-out;\n  transition: transform 0.3s ease-out;\n  transition: transform 0.3s ease-out, -webkit-transform 0.3s ease-out;\n  -webkit-transform: translate(0, -50px);\n  transform: translate(0, -50px);\n}\n\n@media (prefers-reduced-motion: reduce) {\n  .modal.fade .modal-dialog {\n    transition: none;\n  }\n}\n\n.modal.show .modal-dialog {\n  -webkit-transform: none;\n  transform: none;\n}\n\n.modal.modal-static .modal-dialog {\n ...

none;\n}\n\n.modal.fade .modal-dialog {\n  transition: -webkit-transform 0.3s ease-out;\n  transition: transform 0.3s ease-out;\n  transition: transform 0.3s ease-out, -webkit-transform 0.3s ease-out;\n  -webkit-transform: translate(0, -50px);\n  transform: translate(0, -50px);\n}\n\n@media (prefers-reduced-motion: reduce) {\n  .modal.fade .modal-dialog {\n    transition: none;\n  }\n}\n\n.modal.show .modal-dialog {\n  -webkit-transform: none;\n  transform: none;\n}\n\n.modal.modal-static .modal-dialog {\n ...

*   Add `tf.contrib.bayesflow.metropolis_hastings`.
*   Add `log_rate` parameter to `tf.contrib.distributions.Poisson`.
*   Extend `tf.contrib.distributions.bijector` API to handle some non-injective
    transforms.
*   Java:
    *   Generics (e.g., `Tensor<Integer>`) for improved type-safety (courtesy
        @andrewcmyers).
    *   Support for multi-dimensional string tensors.