| f22b9bf | Jean Kossaifi | 09 September 2020, 15:02:28 UTC | Merge pull request #192 from aarmey/patch-1 Update api.rst | 09 September 2020, 15:02:28 UTC |
| 31cc9b4 | Aaron Meyer | 08 September 2020, 20:03:35 UTC | Update api.rst Small spelling fix | 08 September 2020, 20:03:35 UTC |
| 34fdee5 | Jean Kossaifi | 28 August 2020, 14:57:58 UTC | MXNet: fix issue with stacking scalars | 28 August 2020, 15:24:25 UTC |
| 8db27a2 | Jean Kossaifi | 28 August 2020, 14:23:38 UTC | Add symmetric sparse parafac to doc | 28 August 2020, 14:27:39 UTC |
| d5e9eb8 | Jean Kossaifi | 28 August 2020, 14:11:41 UTC | Symmetric parafac with sparse tensors | 28 August 2020, 14:11:41 UTC |
| 489fe7d | Jean Kossaifi | 21 August 2020, 18:33:47 UTC | FIX sparse test. | 21 August 2020, 18:33:47 UTC |
| 70d16b0 | Jean Kossaifi | 21 August 2020, 07:45:42 UTC | Remove forgotten param | 21 August 2020, 07:45:42 UTC |
| bb6a0c5 | Jean Kossaifi | 06 August 2020, 10:40:53 UTC | FIX for Tensorflow, use index_update | 06 August 2020, 10:56:52 UTC |
| 051c262 | Jean Kossaifi | 06 August 2020, 09:57:24 UTC | Add Symmetric Parafac via Robust Tensor Power Iter | 06 August 2020, 10:21:39 UTC |
| 4d96f8c | Jean Kossaifi | 06 August 2020, 09:56:58 UTC | CP: normalize during optimization | 06 August 2020, 09:56:58 UTC |
| 26189bf | Jean Kossaifi | 11 July 2020, 23:34:24 UTC | FIX tensor_dot test for MXNet MXNet currently doesn't support transpose for tensors of order higher than 6 | 11 July 2020, 23:34:24 UTC |
| 9920faa | Jean Kossaifi | 11 July 2020, 20:26:58 UTC | Merge branch 'master' of https://github.com/tensorly/tensorly | 11 July 2020, 20:26:58 UTC |
| 5c85977 | Jean Kossaifi | 11 July 2020, 17:01:41 UTC | Update doc | 11 July 2020, 17:01:41 UTC |
| aa716b8 | Jean Kossaifi | 11 July 2020, 16:56:56 UTC | Adds tensor_dot + batched version and HOM. | 11 July 2020, 16:56:56 UTC |
| 0fa69c7 | Jean Kossaifi | 10 July 2020, 22:32:39 UTC | Merge pull request #185 from aarmey/missing-SVD Using missing values with SVD initialization in PARAFAC | 10 July 2020, 22:32:39 UTC |
| 5f07811 | Aaron Meyer | 10 July 2020, 22:16:18 UTC | Use kruskal_normalise() | 10 July 2020, 22:16:18 UTC |
| eb38212 | Aaron Meyer | 10 July 2020, 20:14:48 UTC | Update candecomp_parafac.py | 10 July 2020, 20:14:48 UTC |
| d5ae55f | Aaron Meyer | 10 July 2020, 20:13:20 UTC | Update candecomp_parafac.py | 10 July 2020, 20:13:20 UTC |
| 190de02 | Aaron Meyer | 10 July 2020, 18:31:19 UTC | Update candecomp_parafac.py | 10 July 2020, 18:31:19 UTC |
| cb099d9 | Jean Kossaifi | 10 July 2020, 18:09:11 UTC | Merge pull request #186 from aarmey/line-search-demo Line search example script | 10 July 2020, 18:09:11 UTC |
| e0fe41c | Jean Kossaifi | 10 July 2020, 18:08:25 UTC | Merge branch 'master' into missing-SVD | 10 July 2020, 18:08:25 UTC |
| 981e8a7 | Aaron Meyer | 10 July 2020, 18:00:02 UTC | Should now initialize weights. | 10 July 2020, 18:00:02 UTC |
| 93ee05b | Aaron Meyer | 10 July 2020, 17:36:45 UTC | Some edits to polish the demo script. | 10 July 2020, 17:36:45 UTC |
| dc422fe | Jean Kossaifi | 10 July 2020, 00:38:49 UTC | Tenalg Backend: correctly set global tenalg_backend (when local_threadsafe=False). | 10 July 2020, 00:38:49 UTC |
| 7b8a6c3 | Jean Kossaifi | 10 July 2020, 00:37:37 UTC | Use default moveaxis with sparse | 10 July 2020, 00:37:37 UTC |
| e8546a2 | Jean Kossaifi | 10 July 2020, 00:36:09 UTC | Default implementation for moveaxis | 10 July 2020, 00:36:09 UTC |
| 03a51c5 | Jean Kossaifi | 09 July 2020, 12:05:52 UTC | Merge pull request #183 from aarmey/line-search PARAFAC line search | 09 July 2020, 12:05:52 UTC |
| 817002f | Aaron Meyer | 08 July 2020, 18:37:05 UTC | Add file. | 08 July 2020, 18:37:05 UTC |
| 0da9e4e | Aaron Meyer | 08 July 2020, 18:01:05 UTC | Add check that error monotonically decreases. | 08 July 2020, 18:01:05 UTC |
| 4c6a2ef | Aaron Meyer | 08 July 2020, 17:50:11 UTC | Fixed typo. | 08 July 2020, 17:50:11 UTC |
| db4be41 | Aaron Meyer | 08 July 2020, 15:16:37 UTC | Acceleration adjustments from the N-way Toolbox | 08 July 2020, 15:16:37 UTC |
| b2fd9ec | Aaron Meyer | 07 July 2020, 14:55:58 UTC | Compare error to last iteration, so that we only calculate the error once unless line search fails. | 07 July 2020, 14:55:58 UTC |
| df40fa9 | Aaron Meyer | 07 July 2020, 14:32:52 UTC | Linting. | 07 July 2020, 14:32:52 UTC |
| 2d95da7 | Aaron Meyer | 07 July 2020, 14:31:45 UTC | Added documentation for error_calc() | 07 July 2020, 14:31:45 UTC |
| 53652a4 | Aaron Meyer | 07 July 2020, 14:19:09 UTC | Changed to Bro line search implementation. | 07 July 2020, 14:19:09 UTC |
| 31fcc9d | Aaron Meyer | 07 July 2020, 13:34:46 UTC | Fix missing jump scaling. | 07 July 2020, 13:34:46 UTC |
| 387dc55 | Aaron Meyer | 05 July 2020, 17:10:54 UTC | Fixed Jax backend. | 05 July 2020, 17:10:54 UTC |
| 570c51f | Aaron Meyer | 05 July 2020, 16:59:20 UTC | Working for PARAFAC. | 05 July 2020, 16:59:20 UTC |
| f66a8f9 | Aaron Meyer | 04 July 2020, 23:10:03 UTC | Performance now better. Included benchmarking workbook. | 04 July 2020, 23:10:03 UTC |
| a86c45d | Aaron Meyer | 03 July 2020, 22:46:22 UTC | Now works with fixed modes. | 03 July 2020, 22:46:22 UTC |
| 118e7ce | Aaron Meyer | 03 July 2020, 20:51:45 UTC | Unify error calculation and fix necessary tensor copy. | 03 July 2020, 20:51:45 UTC |
| 303a8f6 | Jean Kossaifi | 03 July 2020, 11:36:01 UTC | Merge branch 'master' into line-search | 03 July 2020, 11:36:01 UTC |
| 0266447 | Aaron Meyer | 03 July 2020, 00:08:31 UTC | Revert "Reduce tensor reconstruction." This reverts commit a776373b6285d7b2ef59dd9f4e8a8909d23cffc0. | 03 July 2020, 00:08:31 UTC |
| a776373 | Aaron Meyer | 03 July 2020, 00:00:26 UTC | Reduce tensor reconstruction. | 03 July 2020, 00:00:26 UTC |
| f999745 | Jean Kossaifi | 02 July 2020, 20:55:15 UTC | Merge pull request #184 from aarmey/testing-fixes Fix testing errors | 02 July 2020, 20:55:15 UTC |
| 819ba2a | Aaron Meyer | 01 July 2020, 15:06:03 UTC | Update test_candecomp_parafac.py | 01 July 2020, 15:06:03 UTC |
| 9e64803 | Aaron Meyer | 01 July 2020, 14:47:32 UTC | Update test_candecomp_parafac.py | 01 July 2020, 14:47:32 UTC |
| 4a66433 | Aaron Meyer | 01 July 2020, 14:44:29 UTC | Fix sporadic masked Tucker errors. | 01 July 2020, 14:44:29 UTC |
| 8987dbc | Aaron Meyer | 01 July 2020, 14:24:11 UTC | Fixed Jax masking shape in kr. | 01 July 2020, 14:24:11 UTC |
| 03ece99 | Aaron Meyer | 01 July 2020, 14:15:38 UTC | Lower rank of MPS to prevent sporadic testing errors. | 01 July 2020, 14:15:38 UTC |
| d10f16b | Aaron Meyer | 01 July 2020, 14:12:49 UTC | Reduce rank of MPS test. | 01 July 2020, 14:12:49 UTC |
| 2b57318 | Aaron Meyer | 01 July 2020, 14:11:40 UTC | Make masked PARAFAC test more reliable. | 01 July 2020, 14:11:40 UTC |
| 3b9dab1 | Aaron Meyer | 01 July 2020, 04:07:37 UTC | Don't need to tl.copy. | 01 July 2020, 04:07:37 UTC |
| 351f62a | Aaron Meyer | 01 July 2020, 03:40:58 UTC | Update candecomp_parafac.py | 01 July 2020, 03:40:58 UTC |
| 5fa4d37 | Aaron Meyer | 30 June 2020, 22:25:08 UTC | Mostly setup. | 30 June 2020, 22:25:08 UTC |
| bf360e2 | Aaron Meyer | 29 June 2020, 21:23:49 UTC | Add line search for dense PARAFAC solving | 29 June 2020, 21:23:49 UTC |
| a1e4e2a | Jean Kossaifi | 22 June 2020, 13:59:02 UTC | Merge pull request #171 from cohenjer/feature_fixed-mode Fixing modes in parafac and nonegative_parafac | 22 June 2020, 13:59:02 UTC |
| e452f3e | COHEN Jeremy | 22 June 2020, 13:42:40 UTC | changed assert condition | 22 June 2020, 13:42:40 UTC |
| 74c2015 | cohenjer | 22 June 2020, 13:34:44 UTC | Update tensorly/decomposition/tests/test_candecomp_parafac.py Co-authored-by: Jean Kossaifi <jean.kossaifi@gmail.com> | 22 June 2020, 13:34:44 UTC |
| 865292a | COHEN Jeremy | 22 June 2020, 13:30:35 UTC | Added unit testing for the fixed_mode feature | 22 June 2020, 13:30:35 UTC |
| 6ec0209 | COHEN Jeremy | 22 June 2020, 12:32:00 UTC | Merge branch 'dev' into feature_fixed-mode | 22 June 2020, 12:32:00 UTC |
| 16eb359 | Jean Kossaifi | 22 June 2020, 12:06:34 UTC | Doc: Add Logo uao | 22 June 2020, 12:06:34 UTC |
| 459b6f6 | Jean Kossaifi | 22 June 2020, 12:06:16 UTC | Sparse Backend: shape is now builtin | 22 June 2020, 12:06:16 UTC |
| f9be446 | Jean Kossaifi | 19 June 2020, 14:37:58 UTC | DOC: removed old authors file and added Athens uni | 19 June 2020, 14:37:58 UTC |
| f72be60 | Jean Kossaifi | 17 June 2020, 21:34:12 UTC | DOC: update authors page | 17 June 2020, 21:34:12 UTC |
| 6bbadf7 | Jean Kossaifi | 15 June 2020, 20:05:09 UTC | Merge pull request #175 from aarmey/PARAFAC2-init Bugfix: PARAFAC2 not using "init" option | 15 June 2020, 20:05:09 UTC |
| ea0acd8 | Jean Kossaifi | 15 June 2020, 20:02:52 UTC | Merge pull request #173 from aarmey/masked-tucker Implement masked Tucker | 15 June 2020, 20:02:52 UTC |
| 63d875b | Aaron Meyer | 15 June 2020, 16:43:52 UTC | Pass in SVD. | 15 June 2020, 16:43:52 UTC |
| 82a06e6 | Aaron Meyer | 15 June 2020, 16:40:38 UTC | Fix init passing. | 15 June 2020, 16:40:38 UTC |
| 6c8ea1e | Aaron Meyer | 13 June 2020, 16:12:19 UTC | Fix tensorflow indexing. | 13 June 2020, 16:12:19 UTC |
| 59d617f | Aaron Meyer | 13 June 2020, 15:59:13 UTC | Add warning about SVD initialization within PARAFAC. | 13 June 2020, 15:59:13 UTC |
| d0ed464 | Aaron Meyer | 13 June 2020, 15:55:22 UTC | Move masking before first iteration. | 13 June 2020, 15:55:22 UTC |
| a2f4c59 | Aaron Meyer | 13 June 2020, 04:01:08 UTC | Fixed. | 13 June 2020, 04:01:08 UTC |
| 5261885 | Aaron Meyer | 13 June 2020, 00:20:18 UTC | Update test_tucker.py | 13 June 2020, 00:20:18 UTC |
| d1c3dd6 | Aaron Meyer | 13 June 2020, 00:02:54 UTC | Update test_tucker.py | 13 June 2020, 00:02:54 UTC |
| d3d7813 | Aaron Meyer | 13 June 2020, 00:01:44 UTC | Added a more stringent test. | 13 June 2020, 00:01:44 UTC |
| 117e3c1 | Aaron Meyer | 11 June 2020, 17:59:04 UTC | Update test_tucker.py | 11 June 2020, 17:59:04 UTC |
| a308446 | Aaron Meyer | 11 June 2020, 17:56:29 UTC | Implement masked Tucker. | 11 June 2020, 17:56:29 UTC |
| b479f18 | Jean Kossaifi | 03 June 2020, 00:25:02 UTC | Merge pull request #150 from MarieRoald/parafac2 PARAFAC2 in TensorLy | 03 June 2020, 00:25:02 UTC |
| e72a43d | Marie Roald | 01 June 2020, 19:02:50 UTC | Merge branch 'master' into parafac2 | 01 June 2020, 19:02:50 UTC |
| 915502e | Marie Roald | 01 June 2020, 19:02:35 UTC | Merge remote-tracking branch 'tensorly/master' | 01 June 2020, 19:02:35 UTC |
| 38a0c3a | Marie Roald | 01 June 2020, 18:43:15 UTC | Reverted to_numpy fix since it was fixed in the backend | 01 June 2020, 18:43:15 UTC |
| 5108303 | Marie Roald | 01 June 2020, 09:30:32 UTC | Casting singleton tensor norm to numpy to avoid dtype issues in TensorFlow | 01 June 2020, 09:30:32 UTC |
| 896f9b6 | Jean Kossaifi | 31 May 2020, 20:32:32 UTC | FIX Tensorflow norm (order 0 tensors) | 31 May 2020, 20:32:32 UTC |
| 589f254 | Yngve Mardal Moe | 31 May 2020, 19:39:51 UTC | Reverted previous change since we couldn't do a tl.sum with a list as input for pytorch and mxnet backends | 31 May 2020, 19:39:51 UTC |
| 1ed6e4c | Yngve Mardal Moe | 31 May 2020, 19:08:25 UTC | Workaround for the weird TensorFlow behaviour that makes (x**2) change the dtype of x from float32 to float64, thus making a division later raise an error. | 31 May 2020, 19:08:25 UTC |
| 4040012 | Yngve Mardal Moe | 31 May 2020, 18:36:21 UTC | Fixed so congruence is computed correctly and adapted to the changed convergence tolerance in PARAFAC2 | 31 May 2020, 18:36:21 UTC |
| 462564a | Marie Roald | 31 May 2020, 18:18:43 UTC | Added missing square in data norm computation (used for convergence criterion) | 31 May 2020, 18:18:43 UTC |
| 137d3c1 | Marie Roald | 31 May 2020, 18:15:10 UTC | Removed unused variable | 31 May 2020, 18:15:10 UTC |
| 654e35d | Marie Roald | 31 May 2020, 18:14:37 UTC | Fixed so the convergence tolerance is based on the change in relative reconstruction error | 31 May 2020, 18:14:37 UTC |
| 277153d | Marie Roald | 31 May 2020, 18:11:22 UTC | Changed docstring for PARAFAC2 to make it clearer that it's for third order tensors and that the second mode changes over the third mode. | 31 May 2020, 18:11:22 UTC |
| e66f658 | COHEN Jeremy | 29 May 2020, 13:25:55 UTC | Added fixed_modes functionality | 29 May 2020, 14:02:25 UTC |
| e4f8127 | Marie Roald | 24 May 2020, 07:49:21 UTC | Removed set backend | 24 May 2020, 07:49:21 UTC |
| 8075792 | Marie Roald | 24 May 2020, 07:37:30 UTC | Changed tolerance and changed from np.ones to tl.ones | 24 May 2020, 07:37:30 UTC |
| 2e29e6b | Marie Roald | 24 May 2020, 07:37:06 UTC | Fixed index_update error | 24 May 2020, 07:37:06 UTC |
| 7f03ad8 | Marie Roald | 24 May 2020, 07:36:58 UTC | Fixed so we no longer loop over a tl.tensor for TF backend | 24 May 2020, 07:36:58 UTC |
| 8d82dbb | Marie Roald | 24 May 2020, 06:49:48 UTC | Fixed numpy ones to tl ones instead and reduced tolerance of normalize factors test | 24 May 2020, 06:49:48 UTC |
| 3be3217 | Marie Roald | 24 May 2020, 06:04:14 UTC | Fixed normalization test of PARAFAC2 to check magnitude | 24 May 2020, 06:04:14 UTC |
| 1de8bfc | Marie Roald | 23 May 2020, 18:48:22 UTC | Changed shapes and rank so all tests use a tensor with unique PARAFAC2 decomposition | 23 May 2020, 18:48:22 UTC |
| 4007751 | Marie Roald | 23 May 2020, 18:40:44 UTC | Fixed issue with normalisation not happening inplace Previously, the factor variable was replaced in a loop. However, that didn't change the factors list where each factor was stored. Now, the factors list is updated instead | 23 May 2020, 18:40:44 UTC |
