e175478 | Rasmus Bonnevie | 02 August 2017, 15:36:50 UTC | fixed issues with changes to TF syntax. | 02 August 2017, 15:36:50 UTC |
4401562 | Rasmus Bonnevie | 31 January 2017, 22:36:41 UTC | Added derivative and Hessian kernel for Stationary kernels | 31 January 2017, 22:36:41 UTC |
691682d | Rasmus Bonnevie | 31 January 2017, 22:36:11 UTC | Added general functionality for block-structured kernels and recursive co-kriging | 31 January 2017, 22:36:11 UTC |
6745ccd | Mark van der Wilk | 31 January 2017, 16:36:17 UTC | TF futureproofing: tf.mul -> tf.multiply. (#325) | 31 January 2017, 16:36:17 UTC |
2906363 | James Hensman | 31 January 2017, 15:58:41 UTC | Notes for contributors and changes to readme (#317) * First shot at notes for contributors and changes to readme * changing headings * Update contributing.md * Update contributing.md notes for version numbering. * losing the link to google python style guide. | 31 January 2017, 15:58:41 UTC |
b4dce8c | Alexander G. de G. Matthews | 31 January 2017, 15:00:40 UTC | Merge pull request #174 from GPflow/tracer Tracer and wrapped session for GPflow | 31 January 2017, 15:00:40 UTC |
c1295e1 | Mark van der Wilk | 29 January 2017, 17:46:30 UTC | Update session.py | 29 January 2017, 17:46:30 UTC |
0935527 | Mark van der Wilk | 29 January 2017, 17:32:58 UTC | Improved coverage. | 29 January 2017, 17:32:58 UTC |
4c7cbed | Mark van der Wilk | 26 January 2017, 18:54:07 UTC | Simplify session.py, extended tests. | 26 January 2017, 18:54:07 UTC |
13d548a | Mark van der Wilk | 26 January 2017, 18:21:02 UTC | Added test, addressed comments. | 26 January 2017, 18:21:02 UTC |
8a4e023 | Mark van der Wilk | 26 January 2017, 17:59:45 UTC | Merge branch 'master' of github.com:GPflow/GPflow into tracer | 26 January 2017, 17:59:45 UTC |
263b2fd | Alexander G. de G. Matthews | 19 January 2017, 10:05:21 UTC | Sgpr notes (#299) * Small corrections to SGPR notes. * adding missing equals signs. * Correcting typo in notebook. * Small tweak to get code cov to run. | 19 January 2017, 10:05:21 UTC |
2e43abe | Mark van der Wilk | 17 January 2017, 15:10:02 UTC | Fixed each_time warning. | 17 January 2017, 15:10:02 UTC |
48452cf | Mark van der Wilk | 13 January 2017, 18:06:47 UTC | Modified existing profiling tools to be configurable through settings. - Tracer child class is now not abstract any more. - `get_session()` now returns a constructed object. | 13 January 2017, 18:08:14 UTC |
0c48a3f | Mark van der Wilk | 13 January 2017, 16:39:44 UTC | Merge branch 'master' of github.com:GPflow/GPflow into tracer | 13 January 2017, 16:39:44 UTC |
7b904f4 | Mark van der Wilk | 13 January 2017, 16:03:40 UTC | Reformatting code for PEP8. | 13 January 2017, 16:03:40 UTC |
6add8d9 | Mark van der Wilk | 13 January 2017, 16:03:27 UTC | Merge branch 'master' of github.com:GPflow/GPflow into tracer | 13 January 2017, 16:03:27 UTC |
3863e13 | James Hensman | 13 January 2017, 15:26:26 UTC | updating tf version in readthedocs (#319) | 13 January 2017, 15:26:26 UTC |
b970ee6 | Alexis Boukouvalas | 13 January 2017, 14:49:06 UTC | link GPLVM documentation (#318) * link GPLVM documentation * relative links to notebooks * moving stray notebook location | 13 January 2017, 14:49:06 UTC |
0e4b3c4 | Mark van der Wilk | 13 January 2017, 12:25:25 UTC | Compatibility with next version of TensorFlow (#316) * Compatibility with next version of TensorFlow. - Remove references to batch_matmul. - Various functions that need to be renamed. | 13 January 2017, 12:25:25 UTC |
252baea | James Hensman | 12 January 2017, 11:39:17 UTC | Update README.md (#311) | 12 January 2017, 11:39:17 UTC |
bba8e96 | James Hensman | 11 January 2017, 16:18:25 UTC | Bonnevie quadrature refact (#310) * Refactored all quadrature functions into a separate file and updated all existing functions to use it * ekernels.py made compatible with quadrature.py * removal of quadcov function * removal of quadcov function * resolve merge conflict. * made code 2.7 compatible * Small cleanup. | 11 January 2017, 16:18:25 UTC |
b13937d | Mark van der Wilk | 11 January 2017, 09:06:53 UTC | Changes for tf0.12 (#307) * Changes for tf0.12. * Modified tests as well, updated readme & version. * Update .travis.yml | 11 January 2017, 09:06:53 UTC |
2049274 | James Hensman | 20 December 2016, 15:02:54 UTC | adding overflow check and bugfix. (#303) | 20 December 2016, 15:02:54 UTC |
05042f4 | Alexander G. de G. Matthews | 14 December 2016, 13:45:09 UTC | Sgpr notes (#296) * Small corrections to SGPR notes. * adding missing equals signs. | 14 December 2016, 13:45:09 UTC |
851fac3 | James Hensman | 05 December 2016, 20:43:12 UTC | minor updates to hmc.py (#293) | 05 December 2016, 20:43:12 UTC |
7ae7fff | James Hensman | 05 December 2016, 09:28:03 UTC | fix minor bug in MultiClass when passing invlink (#291) | 05 December 2016, 09:28:03 UTC |
9366211 | Alexander G. de G. Matthews | 30 November 2016, 08:09:48 UTC | Improve test coverage for stochastic gradients with tf optimizer. (#287) * Adding sequential generation to minibatch class. * Test minibatch data. * Correction on basis of test. * Pulling out sequence generation. * Building up svgp regression test. * Pulling out minibatch functionality into new python file. * adding test of sampler logic. | 30 November 2016, 08:09:48 UTC |
262f652 | Alexander G. de G. Matthews | 26 November 2016, 15:06:40 UTC | Feed dict needs to be updated on each iteration. (#285) | 26 November 2016, 15:06:40 UTC |
d53a9c6 | James Hensman | 26 November 2016, 12:43:59 UTC | Paramlist constructor bugfix (#282) * changed default argument behaviour for paramlist * tests to cover paramlist constructor | 26 November 2016, 12:43:59 UTC |
e6579d4 | Hugh Salimbeni | 25 November 2016, 17:18:13 UTC | added test for multiple Y columns. Fix in SGPR and GPRFITC (#283) | 25 November 2016, 17:18:13 UTC |
3ab7139 | James Hensman | 24 November 2016, 14:44:36 UTC | fixing bug #277, copy error in mean function constructors (#278) * fixing bug #277, copy error in mean functino constructors * improved default logic for mean fnction constructors | 24 November 2016, 14:44:36 UTC |
1b7b2df | Alexander G. de G. Matthews | 24 November 2016, 10:46:50 UTC | Merge pull request #276 from GPflow/update-travis Update travis | 24 November 2016, 10:46:50 UTC |
c0bfa30 | James Hensman | 24 November 2016, 10:29:24 UTC | Merge branch 'master' into update-travis | 24 November 2016, 10:29:24 UTC |
876a511 | James Hensman | 24 November 2016, 10:12:40 UTC | travis now uses relase version of tf | 24 November 2016, 10:12:40 UTC |
1de5083 | James Hensman | 24 November 2016, 09:59:19 UTC | docs now use python 3.4 (#275) | 24 November 2016, 09:59:19 UTC |
6f036af | James Hensman | 24 November 2016, 09:40:30 UTC | trying puthon 3.4 | 24 November 2016, 09:40:30 UTC |
db38323 | Mark van der Wilk | 23 November 2016, 18:31:25 UTC | Kernel expectations (#195) This squash-commit contains a large volume of work by @markvdw and @alexisboukouvalas. I'm keeping all of the commit history here for posterity. Interested viewers can see some discussion on github, under pull request #195. * Initial code for new kernel expectations. - RBF done. - Added another transformation in etransforms.py. * first step to merge gplvm and kernexp approaches * Added kernel expectations of linear, and eKxz for RBF. - NB: Linear still need to be tested better. - Todo: exKxz for linear. * Added multidimensional quadrature in `Kern` base class for kernel expectations. - Replaced monte carlo tests with more reliable quadrature tests. - Added exKxz for `Linear` kernel * testing new ekernels code, starting on active dimensions, further testing needed as well as modifying Bayesian GPLVM code * Linear and Polynomial kernels did not respect active_dims properly in Kdiag. - Added slice call to Linear.Kdiag - Fixed TestSlice to test more kernels. - Fxied TestSlice to have the correct inputdim. * Begin work on sum kernel, smoothing out active_dims for ekernels. * Fixed `input_dim` in `test_kerns.py`. Added assertion. * kernexp quadrature now works with `active_dims`. - exKzx which doesn't work now raises an error from TensorFlow. - Various other assertions. * Better deduction of `input_dim` for `kernels.Combination`. * Small fix of test. * Fixed issue of KzxKxz in Add kernel. Solution checks for diagonal q(X) and performs quadrature on the covariance of KzxKxz if not diagonal. * GPLVM now works with new kernel expectation code. - Added a new DiagMatrix transform. - Removed legacy code. - Modified GPLVM to accept full covariance matrices. * Fixing a small error in DiagMatrix transform. * Added warnings. * Removed etransforms code. BlockTriDiagonalTransform is now only used in tests. * Prevent `TridiagonalBlockRep` from being tested as a `Transform`. * improvements to DiagonalMatrix transform * improved testing of kernel slice * update test to use kernels * `ekernels.RBF` ARD bug fixed. * Increased test coverage. * Increased test coverage. * docstring for gplvm, removed unused variable in ekernels * testing composite kernels in gplvm * Added Prod to `ekernels.py`. * Fixed usage of `tf.gather_nd`, which does not have a gradient in `_slice_cov()`. * add GPLVM notebook, increase testing to include Prod kernels, add documentation stub, fix bibliography * improve GPLVM notebook with working example * Fixed bug that `X_var` in `BayesianGPLVM` gets slightly different values. * Quadrature can be switched off, plus the appropriate checks. * Kernel expectations now accept 2D variances. * Fix to test. * Reworked quadrature code in Add in anticipation of adding exact expectations for certain pairs of kernels. * Added Linear + Add cross terms for overlapping active_dims. * Initial try for extra test for `ekernels.Add` cross terms. * `_slice_cov` now again compatible with numpy arrays. * Requested code reviews. | 23 November 2016, 18:31:25 UTC |
419c43b | Alexander G. de G. Matthews | 21 November 2016, 10:45:35 UTC | Merge pull request #273 from GPflow/jameshensman-readme-patch point to tensorflow 0.11 in readme (not 0.10) | 21 November 2016, 10:45:35 UTC |
f2883d5 | James Hensman | 21 November 2016, 09:08:09 UTC | Merge branch 'master' into jameshensman-readme-patch | 21 November 2016, 09:08:09 UTC |
9999cc4 | James Hensman | 21 November 2016, 09:07:58 UTC | point to tensorflow 0.11 in readme (not 0.10) | 21 November 2016, 09:07:58 UTC |
efbea03 | Keisuke Fujii | 20 November 2016, 08:19:13 UTC | Patch setattr (#272) * Patch for #269 * Replacement of Parameterized now raises _needs_recompile flag. * Fix typo | 20 November 2016, 08:19:13 UTC |
4005d83 | David Sheldon | 18 November 2016, 15:04:28 UTC | Update kernels.ipynb (#271) Corrected a couple of typos in the text. | 18 November 2016, 15:04:28 UTC |
9de24ca | Keisuke Fujii | 17 November 2016, 12:33:17 UTC | Patch for #269 (#270) Fix bug in equality comparison: fails on some systems due to floating point error. | 17 November 2016, 12:33:17 UTC |
3f9c248 | Alexander G. de G. Matthews | 09 November 2016, 14:12:11 UTC | Bibtex docs (#263) * Adding the GPflow bibtex. * Tweaking bibtex * Tweaking Readme order. | 09 November 2016, 14:12:11 UTC |
c84bc39 | Alexander G. de G. Matthews | 07 November 2016, 16:10:42 UTC | Merge pull request #145 from GPflow/multi-likelihood Multi likelihood | 07 November 2016, 16:10:42 UTC |
7131e7f | Alexander G. de G. Matthews | 07 November 2016, 13:21:14 UTC | Merge branch 'master' into multi-likelihood | 07 November 2016, 13:21:14 UTC |
a1c5340 | James Hensman | 04 November 2016, 17:16:23 UTC | add fn to gte the length of a paramlist (#220) * add fn to ge the length of a paramlist * a very simple test for the length of a paramlist | 04 November 2016, 17:16:23 UTC |
d0b3a7e | James Hensman | 04 November 2016, 14:51:45 UTC | small update to meanfunction tests | 04 November 2016, 14:51:45 UTC |
2500a54 | James Hensman | 04 November 2016, 14:23:25 UTC | Merge branch 'master' into multi-likelihood | 04 November 2016, 14:23:25 UTC |
2a755d5 | Alexander G. de G. Matthews | 04 November 2016, 14:23:00 UTC | Incrementing minor version number to 0.3.4 (#262) * Incrementing minor version number. * Correcting typo. | 04 November 2016, 14:23:00 UTC |
7ee159d | Alexander G. de G. Matthews | 04 November 2016, 13:38:14 UTC | Merge pull request #261 from GPflow/autoflow-fix Autoflow fix | 04 November 2016, 13:38:14 UTC |
bbb0c6c | Alexander G. de G. Matthews | 04 November 2016, 10:37:08 UTC | Merge branch 'master' into autoflow-fix | 04 November 2016, 10:37:08 UTC |
dd4727f | Mark van der Wilk | 04 November 2016, 08:24:25 UTC | Better deduction of `input_dim` for `kernels.Combination` (#258) * Better deduction of `input_dim` for `kernels.Combination`. * Fix small off-by-one-error k.active_dims contain indices, and we want number of columns. | 04 November 2016, 08:24:25 UTC |
7bdb7f8 | James Hensman | 03 November 2016, 22:09:07 UTC | minor edits to tests for new update_feed_dict functionality | 03 November 2016, 22:09:07 UTC |
834290e | James Hensman | 03 November 2016, 21:04:50 UTC | small change to the order of operations | 03 November 2016, 21:04:50 UTC |
7cacea5 | James Hensman | 03 November 2016, 21:00:43 UTC | Merge branch 'autoflow-fix' of github.com:gpflow/GPflow into autoflow-fix | 03 November 2016, 21:00:43 UTC |
5eea975 | James Hensman | 03 November 2016, 21:00:32 UTC | minor bugfix | 03 November 2016, 21:00:32 UTC |
a69f023 | James Hensman | 03 November 2016, 20:55:13 UTC | Merge branch 'master' into autoflow-fix | 03 November 2016, 20:55:13 UTC |
597eb6e | James Hensman | 03 November 2016, 20:49:45 UTC | changes to the get_feed_dict mechanism | 03 November 2016, 20:49:45 UTC |
d7dc9e8 | James Hensman | 03 November 2016, 20:11:06 UTC | aadint test to cover bad AutoFlow computations | 03 November 2016, 20:11:06 UTC |
377a45b | Alexander G. de G. Matthews | 03 November 2016, 20:01:20 UTC | Updates to readme file. (#259) | 03 November 2016, 20:01:20 UTC |
7ff195a | James Hensman | 03 November 2016, 14:24:46 UTC | bugfix in vgp (for multiple concurrent Y) | 03 November 2016, 14:24:46 UTC |
f3ecd45 | James Hensman | 03 November 2016, 14:00:16 UTC | Merge branch 'multi-likelihood' of github.com:gpflow/GPflow into multi-likelihood | 03 November 2016, 14:00:16 UTC |
ac6f42e | James Hensman | 03 November 2016, 14:00:04 UTC | Merge remote-tracking branch 'origin/master' into multi-likelihood | 03 November 2016, 14:00:04 UTC |
f1ec927 | James Hensman | 03 November 2016, 13:59:54 UTC | Merge branch 'master' into multi-likelihood | 03 November 2016, 13:59:54 UTC |
8e1d014 | Alexander G. de G. Matthews | 03 November 2016, 11:39:45 UTC | Correcting a typo in readme. (#255) | 03 November 2016, 11:39:45 UTC |
55e5bbe | Alexander G. de G. Matthews | 03 November 2016, 11:18:05 UTC | Updating list of contributors (#254) * Updating list of contributors. * Adding Ivo. | 03 November 2016, 11:18:05 UTC |
d5addce | Mark van der Wilk | 02 November 2016, 12:33:01 UTC | active_dims fix in Kdiag (#253) * Linear and Polynomial kernels did not respect active_dims properly in Kdiag. - Added slice call to Linear.Kdiag - Fixed TestSlice to test more kernels. - Fxied TestSlice to have the correct inputdim. * Fixed `input_dim` in `test_kerns.py`. Added assertion. | 02 November 2016, 12:33:01 UTC |
7801f75 | James Hensman | 01 November 2016, 17:36:32 UTC | merging test_likelihoods.py | 01 November 2016, 17:36:32 UTC |
82af806 | James Hensman | 01 November 2016, 14:06:04 UTC | Merge branch 'master' into multi-likelihood | 01 November 2016, 14:06:04 UTC |
3ba0758 | James Hensman | 01 November 2016, 14:05:56 UTC | merging | 01 November 2016, 14:05:56 UTC |
f66dda5 | James Hensman | 01 November 2016, 12:03:52 UTC | improvements to doctring of kernels regarding slices (#252) | 01 November 2016, 12:03:52 UTC |
4311448 | James Hensman | 01 November 2016, 11:07:32 UTC | improvements to setup.py (#251) | 01 November 2016, 11:07:32 UTC |
94a756d | James Hensman | 01 November 2016, 10:47:26 UTC | make sure log jacobian is computed even for fixed variables (#248) | 01 November 2016, 10:47:26 UTC |
97bda5a | icouckuy | 31 October 2016, 18:06:07 UTC | Fix matpackops compilation with gcc 6 (#222) | 31 October 2016, 18:06:06 UTC |
b8b91e0 | Alexander G. de G. Matthews | 31 October 2016, 09:54:39 UTC | Merge pull request #239 from GPflow/graphing Graphing | 31 October 2016, 09:54:39 UTC |
70eae34 | Alexander G. de G. Matthews | 28 October 2016, 15:58:02 UTC | Merge branch 'master' into graphing | 28 October 2016, 15:58:02 UTC |
b89794d | Alexander G. de G. Matthews | 28 October 2016, 13:09:15 UTC | Merge pull request #245 from GPflow/version_fix fixing tf version in setup.py | 28 October 2016, 13:09:15 UTC |
23bc64c | James Hensman | 28 October 2016, 12:51:29 UTC | fixing tf version in setup.py | 28 October 2016, 12:51:29 UTC |
bbd58e5 | Alexander G. de G. Matthews | 27 October 2016, 15:20:42 UTC | Merge branch 'master' into graphing | 27 October 2016, 15:20:42 UTC |
512007d | Alexander G. de G. Matthews | 27 October 2016, 15:06:06 UTC | v0.3.3 paper submission. (#243) | 27 October 2016, 15:06:06 UTC |
66508ff | Alexander G. de G. Matthews | 27 October 2016, 14:44:28 UTC | Merge pull request #242 from GPflow/doc_update updating tf version in docs_require | 27 October 2016, 14:44:28 UTC |
3fd0fcf | James Hensman | 27 October 2016, 14:28:45 UTC | Merge branch 'master' into doc_update | 27 October 2016, 14:28:45 UTC |
fbde5b3 | Alexander G. de G. Matthews | 27 October 2016, 14:27:26 UTC | Adding Apache license to all testing files. (#240) | 27 October 2016, 14:27:26 UTC |
ee85279 | Alexander G. de G. Matthews | 27 October 2016, 14:12:57 UTC | Merge pull request #241 from GPflow/tidy_notebooks tidying notebook files | 27 October 2016, 14:12:57 UTC |
d3f62da | James Hensman | 27 October 2016, 14:09:03 UTC | updating tf version in docs_require | 27 October 2016, 14:09:03 UTC |
12d2c80 | James Hensman | 27 October 2016, 14:00:00 UTC | moving notebooks around | 27 October 2016, 14:00:00 UTC |
5c523ec | James Hensman | 26 October 2016, 21:56:37 UTC | small fixes to scope tests | 26 October 2016, 21:56:37 UTC |
f08193d | James Hensman | 26 October 2016, 21:24:07 UTC | fixes in test_variational | 26 October 2016, 21:24:07 UTC |
e5b4571 | James Hensman | 26 October 2016, 21:15:26 UTC | fixing build_prior_KL test | 26 October 2016, 21:15:26 UTC |
cf4b659 | James Hensman | 26 October 2016, 21:03:30 UTC | pep8ifying test_variational | 26 October 2016, 21:03:30 UTC |
27addff | James Hensman | 26 October 2016, 20:27:16 UTC | improvements to model picklability | 26 October 2016, 20:27:16 UTC |
a6b6b42 | James Hensman | 26 October 2016, 20:15:35 UTC | model.py: make _objective functn on its own graph | 26 October 2016, 20:15:35 UTC |
5a2ea6e | James Hensman | 26 October 2016, 20:02:23 UTC | some small changes to get placeholders off the default graph | 26 October 2016, 20:02:23 UTC |
1ebcba3 | James Hensman | 26 October 2016, 19:51:50 UTC | simplification of Logistic transform | 26 October 2016, 19:51:50 UTC |
ef93b12 | James Hensman | 26 October 2016, 19:26:32 UTC | minor fixes in testDataHolder | 26 October 2016, 19:26:32 UTC |
4cef013 | James Hensman | 26 October 2016, 19:15:26 UTC | chanegs to the way fixed placeholder work Fixed variables are treated as placeholders. In this change, the placeholder is created on the call to make_tf_array, rather than at the moment that the fixed bool is set. This lets the placeholder be created inside the graph context. | 26 October 2016, 19:15:26 UTC |
f5b94dd | James Hensman | 26 October 2016, 19:10:25 UTC | Keep AutoFlow computations off the default graph The Autoflow function now makes its own graph for each autoflow in the model. Some extra tests in test_AutoFlow are added to make sure that it works with dataholders. The dataHolder class has changed a little: it's included in make_tf_array now, so that the placeholders can be made inside the Autoflow graph: before, all placeholders were on the default graph. | 26 October 2016, 19:10:25 UTC |
b811d4f | Alexander G. de G. Matthews | 24 October 2016, 14:39:04 UTC | Merge pull request #238 from GPflow/versionupdate3.2 Versionupdate3.2 | 24 October 2016, 14:39:04 UTC |