ed3e6c3 | Curtis "Fjord" Hawthorne | 29 June 2018, 23:41:59 UTC | bump version (#1220) | 29 June 2018, 23:41:59 UTC |
b8112d8 | Ian Simon | 29 June 2018, 22:56:44 UTC | support cuDNN in events_rnn_graph.py (#1219) * support cuDNN in events_rnn_graph.py * use CudnnCompatibleLSTM for generation | 29 June 2018, 22:56:44 UTC |
86a92c5 | Ian Simon | 29 June 2018, 03:09:31 UTC | support residual connections in events_rnn_graph.py (#1217) * support residual connections in events_rnn_graph.py * specify default hyperparameters in events_rnn_model.py | 29 June 2018, 03:09:31 UTC |
2c3ae9b | Curtis "Fjord" Hawthorne | 27 June 2018, 01:47:36 UTC | Velocity model fixes (#1215) | 27 June 2018, 01:47:36 UTC |
e376d83 | czhuang | 23 June 2018, 00:07:37 UTC | Dilation rate (#1212) * reduce dilation rates and add control * fix lint errors * rewrite max and min syntax * set joblib version to be less than 0.12 since the latest release is failing * fix quotes for test.sh | 23 June 2018, 00:07:37 UTC |
74f16be | czhuang | 19 June 2018, 22:11:28 UTC | Adding dilated convs to Coconet. (#1208) * add dilated option * fix lint errors | 19 June 2018, 22:11:28 UTC |
bf6c8e2 | Curtis "Fjord" Hawthorne | 18 June 2018, 21:29:56 UTC | Add unit test for coconet (#1207) * unit test * add test * lint * lint * lint * py3 | 18 June 2018, 21:29:56 UTC |
a04f5ad | Jesse Engel | 15 June 2018, 20:18:53 UTC | Fix for creating batches where no padding is required. (#1206) | 15 June 2018, 20:18:53 UTC |
c42fae1 | Paige Bailey | 10 June 2018, 14:53:30 UTC | Making "NOTE:"'s consistent in format. (#1201) thanks | 10 June 2018, 14:53:30 UTC |
710519d | Jacob Howcroft | 07 June 2018, 23:57:29 UTC | Coconet: Add strip_default_attrs to SavedModel. (#1200) | 07 June 2018, 23:57:29 UTC |
92399da | Jacob Howcroft | 05 June 2018, 20:43:38 UTC | Coconet: Remove outer_masks from input and fix a couple of errors. (#1197) * Coconet: Remove outer_masks from input and fix a couple of errors. * Coconet: only tile masks if it doesn't match input prolls (assume its 1) * Coconet: Spelled lambda wrong | 05 June 2018, 20:43:38 UTC |
ab6a37b | Ian Simon | 05 June 2018, 01:06:47 UTC | support limited pitch range in multitrack MusicVAE (#1196) * support limited pitch range in multitrack MusicVAE * remove unused constant | 05 June 2018, 01:06:47 UTC |
06f0d2d | Curtis "Fjord" Hawthorne | 04 June 2018, 23:45:57 UTC | Add velocity transcription (#1195) * Add velocity transcription, standalone script. * fix * bump tf api | 04 June 2018, 23:45:57 UTC |
c9962ca | Jacob Howcroft | 01 June 2018, 23:15:14 UTC | Coconet: Add optional hard override for total_gibbs_steps. (#1193) | 01 June 2018, 23:15:14 UTC |
89e4f36 | Jacob Howcroft | 01 June 2018, 22:57:22 UTC | Coconet: Add optional input overrides. (#1192) * Coconet: Add ability to override inputs. * Coconet: Pass overrided inputs to completion masker. * Fix inconsitent quote. | 01 June 2018, 22:57:22 UTC |
75a3d48 | Curtis "Fjord" Hawthorne | 01 June 2018, 22:08:50 UTC | Bump version, clarifications, lint (#1191) * Bump version, fixes. - Clarify dev environment install instructions. - Lint fixes. - Bump version. * allow long lines, sort | 01 June 2018, 22:08:50 UTC |
760262d | czhuang | 01 June 2018, 21:50:50 UTC | Completion masker (#1190) * move completion masker into the graph * remove lib_sampling dependency in lib_tfsampling.py * add default value zeros to outer_masks which will trigger completion mask, fix argument passed into completion mask should be pianorolls * add doc string to property outer_masks * fix typo in doc | 01 June 2018, 21:50:50 UTC |
7ff8868 | Brian Jaress | 31 May 2018, 04:00:10 UTC | Fix for issue 1148 (#1189) * Fix for issue 1148 * Includes arbitrary stylization scripts in pip build * Console_entry_point shim Now that the arbitrary stylization scripts are included, I realized that they also need a small shim to be able to actually run. | 31 May 2018, 04:00:10 UTC |
55692f5 | Curtis "Fjord" Hawthorne | 23 May 2018, 21:05:21 UTC | save a checkpoint every hour during training (#1185) | 23 May 2018, 21:05:21 UTC |
c65a05a | Curtis "Fjord" Hawthorne | 23 May 2018, 20:53:00 UTC | Support float samples in wav_data_to_samples (#1184) * support float samples in wav_data_to_samples * lint | 23 May 2018, 20:53:00 UTC |
923f4d0 | Jacob Howcroft | 21 May 2018, 20:23:30 UTC | Coconet: Switch input and output pianorolls to bool (from float32) (#1181) * Coconet: Switch input and output pianorolls to bool (from float32). I can successfully export and use a SavedModel with this setup. This should lower bandwidth and memory footprint of tensor protos. This doesn't change the main graph, only the sampling graph. It just converts the input from bool to float right away, and vice versa for output. * Delete DS_Stores. Forgot that removing them from the commit comment doesn't actually unstage them. | 21 May 2018, 20:23:30 UTC |
204c828 | Vida Vakilotojar | 18 May 2018, 02:39:12 UTC | Modulo encoding (#1179) * Adds support for Modulo Encoding to Performance RNN. * Fixes lint errors. * Removes ModuloEventSequenceEncoderDecoder, and moves ModuloPerformanceEventSequenceEncoderDecoder to performance_encoder_decoder.py. * Fixes lint error, line too long. * Simplifies PerformanceModuloEncoding to only encode inputs, and changes ModuloPerformanceEventSequenceEncoderDecoder to use both a PerformanceModuloEncoding and a PerformanceOneHotEncoding. * Fixes Lint error. * Adds support for Modulo Encoding to Performance RNN. * Fixes lint errors. * Removes ModuloEventSequenceEncoderDecoder, and moves ModuloPerformanceEventSequenceEncoderDecoder to performance_encoder_decoder.py. * Fixes lint error, line too long. * Simplifies PerformanceModuloEncoding to only encode inputs, and changes ModuloPerformanceEventSequenceEncoderDecoder to use both a PerformanceModuloEncoding and a PerformanceOneHotEncoding. * Fixes Lint error. * Removed leftover code from performance_encoder_decoder_test. * Changes ModuloPerformanceEventSequenceEncoderDecoder so that it itself instantiates its PerformanceModuloEncoding and PerformanceOneHotEncoding members. * Removes unused import causing lint error. * Cleans up the comments. * Fixed labels_to_num_steps(), and other CL review comments. * Creates tables for modulo-encoding of notes, time-shifts, and velocities. * Adds back missing line in __init__.py * Reverts README file of music_vae. * Adds support for Modulo Encoding to Performance RNN. * Fixes lint errors. * Removes ModuloEventSequenceEncoderDecoder, and moves ModuloPerformanceEventSequenceEncoderDecoder to performance_encoder_decoder.py. * Fixes lint error, line too long. * Simplifies PerformanceModuloEncoding to only encode inputs, and changes ModuloPerformanceEventSequenceEncoderDecoder to use both a PerformanceModuloEncoding and a PerformanceOneHotEncoding. * Fixes Lint error. * Adds support for Modulo Encoding to Performance RNN. * Removes ModuloEventSequenceEncoderDecoder, and moves ModuloPerformanceEventSequenceEncoderDecoder to performance_encoder_decoder.py. * Simplifies PerformanceModuloEncoding to only encode inputs, and changes ModuloPerformanceEventSequenceEncoderDecoder to use both a PerformanceModuloEncoding and a PerformanceOneHotEncoding. * Fixes Lint error. * Removed leftover code from performance_encoder_decoder_test. * Changes ModuloPerformanceEventSequenceEncoderDecoder so that it itself instantiates its PerformanceModuloEncoding and PerformanceOneHotEncoding members. * Removes unused import causing lint error. * Cleans up the comments. * Fixed labels_to_num_steps(), and other CL review comments. * Creates tables for modulo-encoding of notes, time-shifts, and velocities. * Adds back missing line in __init__.py * Reverts README file of music_vae. * Removes.DS_Store files. | 18 May 2018, 02:39:12 UTC |
5bcb971 | Curtis "Fjord" Hawthorne | 15 May 2018, 00:59:15 UTC | Don't require futures when on py3 (#1174) | 15 May 2018, 00:59:15 UTC |
d4d5828 | Curtis "Fjord" Hawthorne | 15 May 2018, 00:38:24 UTC | Add support for BEAT annotation (#1175) * add beat annotation * support for beats when splitting * fix * docs | 15 May 2018, 00:38:24 UTC |
d03f656 | Adam Roberts | 11 May 2018, 20:52:05 UTC | Replace occurences of deprecated condiitonal hparam with z_size. (#1173) * Replace occurences of deprecated condiitonal hparam with z_size. * fix lint * fix lint | 11 May 2018, 20:52:05 UTC |
81b6f67 | Pablo Samuel Castro | 11 May 2018, 01:55:50 UTC | Trained model for hierarchical decoders (#1170) * Updating TrainedModel to be compatible with hierarchical decoders. * Update TrainedModel to work with hierarchical decoders. | 11 May 2018, 01:55:50 UTC |
e8e6678 | czhuang | 10 May 2018, 22:01:57 UTC | Depthwise sep conv for coconet (#1172) * add depthwise_separable_conv * fix line too long | 10 May 2018, 22:01:57 UTC |
5746cd4 | Curtis "Fjord" Hawthorne | 10 May 2018, 21:46:14 UTC | fix line continuation typo (#1171) | 10 May 2018, 21:46:14 UTC |
452904c | Jacob Howcroft | 10 May 2018, 00:04:58 UTC | Cleanup and refactor to support TPU inference (#1169) * Coconet: Cleanup lib_tfsampling and add sampling support to export_saved_model. Uses tf.placeholder_with_default for the scalar params so they can be left out of the SavedModel signature. * Fix import order. * Add missing "lib_saved_model.py". | 10 May 2018, 00:04:58 UTC |
fcd5f19 | Curtis "Fjord" Hawthorne | 08 May 2018, 00:17:51 UTC | Switch to tf.logging instead of print (#1168) * switch to tf.logging instead of print * lint * lint * fix | 08 May 2018, 00:17:51 UTC |
9144bd2 | Curtis "Fjord" Hawthorne | 07 May 2018, 22:58:54 UTC | automatically select checkpoint dir for eval (#1167) | 07 May 2018, 22:58:54 UTC |
99fd414 | Ian Simon | 07 May 2018, 22:36:21 UTC | split note sequences more efficiently (#1166) | 07 May 2018, 22:36:21 UTC |
30039fb | czhuang | 07 May 2018, 21:53:59 UTC | Coconet fix eval lr (#1165) * lr eval is no op * fix eval train_op to no op * lr eval is no op * fix eval train_op to no op | 07 May 2018, 21:53:59 UTC |
93b5ab7 | Ian Simon | 07 May 2018, 17:23:43 UTC | actually run hierarchical data converter tests (#1163) | 07 May 2018, 17:23:43 UTC |
786210f | Curtis "Fjord" Hawthorne | 04 May 2018, 17:30:32 UTC | Update coconet_train internal visibility (#1162) * visibility * lint | 04 May 2018, 17:30:32 UTC |
f4dcbf3 | Adam Roberts | 03 May 2018, 23:08:10 UTC | MusicVAE: Add training hparams to configs, add hierarchical encoder configs, and update README. (#1161) * Add training hparams to configs and update settings. * Add training hparams to configs and update settings. | 03 May 2018, 23:08:10 UTC |
c4337d9 | Jacob Howcroft | 03 May 2018, 17:45:37 UTC | Coconet: Remove conv weights initializer when not training (was causing TPU errors and isn't needed for inference) and remove tf.nn.moments from batchnorm when not training (was a bug, shouldn't have been in inference graph). (#1159) | 03 May 2018, 17:45:37 UTC |
8db770e | Adam Roberts | 03 May 2018, 17:08:06 UTC | Update README.md | 03 May 2018, 17:08:06 UTC |
980188b | Adam Roberts | 03 May 2018, 17:07:22 UTC | Update README with link to magentajs. (#1160) | 03 May 2018, 17:07:22 UTC |
4f3568e | Adam Roberts | 02 May 2018, 23:28:55 UTC | Add documentation and TODO concerning meaning of total_time field in … (#1158) * Add documentation and TODO concerning meaning of total_time field in NoteSequence. * Respond to reviewer comments | 02 May 2018, 23:28:55 UTC |
515ce55 | Adam Roberts | 02 May 2018, 22:02:00 UTC | Fix incorrect string formatting. Disable irrelvant lint message. (#1157) | 02 May 2018, 22:02:00 UTC |
4d8d755 | czhuang | 02 May 2018, 19:20:10 UTC | Adding annealed independent blocked Gibbs sampling in the graph for Coconet. (#1153) * add build file * add tfsample import * coconet_sample changes * add tfsample dependency * fix double of piece_length in description * lib_graph changes * completion masker in lib_sampling now * add lib_tfsampling.py * add build file * add tfsample import * coconet_sample changes * add tfsample dependency * fix double of piece_length in description * lib_graph changes * completion masker in lib_sampling now * add lib_tfsampling.py | 02 May 2018, 19:20:10 UTC |
4a0823a | Curtis "Fjord" Hawthorne | 02 May 2018, 17:29:52 UTC | disable flaky midi_hub_test (#1155) | 02 May 2018, 17:29:52 UTC |
5c807d7 | Ian Simon | 02 May 2018, 16:45:57 UTC | add multi-instrument converter to MusicVAE (#1152) * add multi-instrument converter to MusicVAE * add config for chord-conditioned melody model | 02 May 2018, 16:45:57 UTC |
35377d2 | Ian Simon | 01 May 2018, 17:39:07 UTC | refactor Performance RNN conditioning to make it easier to experiment with new control signals (#1151) * refactor Performance RNN conditioning to make it easier to experiment with new control signals * fix tests and lint errors | 01 May 2018, 17:39:07 UTC |
48a1990 | Dan Ringwalt | 02 April 2018, 19:06:15 UTC | Add //visibility:public to protobuf, for use by external Bazel projects. (#1135) | 02 April 2018, 19:06:15 UTC |
dcc8658 | Adam Roberts | 27 March 2018, 22:03:57 UTC | Remove JS code from main repo now that it is in tensorflow/magenta-js (#1133) * Remove js from main repo * Remove js from main repo * Create README.md * Update README.md * Update and rename README to README.md | 27 March 2018, 22:03:57 UTC |
841b876 | Adam Roberts | 27 March 2018, 19:20:07 UTC | Move MusicVAE.js into magenta/js subdir. (#1132) * Move magenta/models/music_vae/js to magenta/js/music_vae * Update README | 27 March 2018, 19:20:07 UTC |
77fd8d7 | Adam Roberts | 27 March 2018, 18:21:48 UTC | Add model support for trio. (#1131) * Add hierarchical conductor wrapper. * Add hierarchical conductor wrapper. * Add hierarchical encoder. * Add hierarchical encoder. * Add documentation. * Add documentation. * Add model support for trios * merge * fix merge * fix merge * fix merge * fix merge * fix merge * fix merge * Add doc * Add readonly to NUM_SPLITS | 27 March 2018, 18:21:48 UTC |
6cf7437 | Curtis "Fjord" Hawthorne | 27 March 2018, 17:44:32 UTC | support converting back to midi (#1129) | 27 March 2018, 17:44:32 UTC |
98a7324 | Curtis "Fjord" Hawthorne | 27 March 2018, 17:27:29 UTC | melody_rnn javascript (#1125) * starting * continuing * more * works * fixes * arbitrary layers * fixes | 27 March 2018, 17:27:29 UTC |
4995b9d | Tero Parviainen | 27 March 2018, 16:36:05 UTC | Use asynchronous API when converting tensors to data sequences. (#1130) This prevents the JavaScript thread from being blocked during the operation. | 27 March 2018, 16:36:05 UTC |
e126144 | Adam Roberts | 26 March 2018, 21:33:46 UTC | Add support for single-instrument hierarchical/conductor models in MusicVAE.js. (#1124) * Add hierarchical conductor wrapper. * Add hierarchical encoder. * Add documentation. * Switch to using CDN for distributing musicvae.js (#1123) * Add hierarchical conductor wrapper. * Add hierarchical encoder. * Add documentation. * Responded to reviewer comments * Update bundle link | 26 March 2018, 21:33:46 UTC |
7a319f3 | Adam Roberts | 26 March 2018, 19:08:19 UTC | Expand examples path in MusicVAE. Use Melody augmentation by default. (#1128) * Expand examples path in musicvae * Add default melody augmentation. * Fix coconet lint | 26 March 2018, 19:08:19 UTC |
ec53d68 | Curtis "Fjord" Hawthorne | 26 March 2018, 18:45:37 UTC | Revert kokoro OMR deps (#1127) | 26 March 2018, 18:45:37 UTC |
095eebc | folex | 26 March 2018, 18:38:58 UTC | Mention directory change before running bazel test (#753) | 26 March 2018, 18:38:58 UTC |
cbfa8ac | Adam Roberts | 22 March 2018, 22:37:12 UTC | Switch to using CDN for distributing musicvae.js (#1123) | 22 March 2018, 22:37:12 UTC |
0d8b2c5 | Curtis "Fjord" Hawthorne | 22 March 2018, 16:34:33 UTC | v0.0.4 (#1121) | 22 March 2018, 16:34:33 UTC |
ec28001 | Adam Roberts | 22 March 2018, 01:53:51 UTC | Remove global deeplearn variable which breaks clients in unsupported browsers. (#1122) * Remove global deeplearn variable which breaks clients in unsupported browsers. * bump version * Update dist | 22 March 2018, 01:53:51 UTC |
3cafc63 | Jacob Howcroft | 21 March 2018, 18:06:16 UTC | Coconet lib_pianoroll: Replace print with tf.logging. (#1120) | 21 March 2018, 18:06:16 UTC |
39383c0 | Adam Roberts | 20 March 2018, 22:35:12 UTC | Add error message when examples file(s) not found. (#1119) * Add error message when examples file not found. * lint * Fix check | 20 March 2018, 22:35:12 UTC |
1f7827f | Adam Roberts | 20 March 2018, 18:23:49 UTC | Add Python3 support to midi_hub and fix tests. (#1117) * Add Python3 support to midi_hub and fix tests. * pip3 compat * lint * lint | 20 March 2018, 18:23:49 UTC |
dd3432d | Adam Roberts | 19 March 2018, 20:02:07 UTC | Add quantization support to CheckpointLoader. (#1113) * Add quantization support to CheckpointLoader. * lint * Bump package version. | 19 March 2018, 20:02:07 UTC |
b996706 | cclauss | 19 March 2018, 16:28:49 UTC | from six.moves import xrange (#1107) | 19 March 2018, 16:28:49 UTC |
2c20c31 | Curtis "Fjord" Hawthorne | 19 March 2018, 16:22:07 UTC | add notebook dep (#1115) | 19 March 2018, 16:22:07 UTC |
ca73164 | Ian Simon | 16 March 2018, 23:50:51 UTC | account for start step in performance steps property, and add MetricPerformance to magenta.music (#1114) * account for start step in performance steps property * add MetricPerformance to magenta.music * add tests for steps property | 16 March 2018, 23:50:51 UTC |
5727f18 | Curtis "Fjord" Hawthorne | 16 March 2018, 21:31:22 UTC | Add javascript midi->NoteSequence conversion (#1097) * midi_io * add * make linter happy * fixes | 16 March 2018, 21:31:22 UTC |
d488422 | Torin Blankensmith | 16 March 2018, 21:15:02 UTC | Currently working on safari + safari mobile iOS 11 (#1090) | 16 March 2018, 21:15:02 UTC |
a5acfbf | Adam Roberts | 15 March 2018, 19:21:37 UTC | Set default MusicVAE master to '' (#1112) * Set default MusicVAE master to '' * Bump pip version | 15 March 2018, 19:21:37 UTC |
9fe80ac | Adam Roberts | 15 March 2018, 19:07:01 UTC | Update MusicVAE.js README (#1111) * Update README * bump npm version | 15 March 2018, 19:07:01 UTC |
0b4b73b | Adam Roberts | 14 March 2018, 21:29:31 UTC | Add DataConverter spec to include with MusicVAE checkpoints. (#1094) * Add DataConverter spec to include with MusicVAE checkpoints. * Add doc for spec * Move loading from spec into data lib * rebase | 14 March 2018, 21:29:31 UTC |
09ace84 | Adam Roberts | 14 March 2018, 21:21:29 UTC | Add temperature to sampling. (#1108) * Add temperature to sampling. * Version bump to 0.1.0 * Update bundle * Version bump to 1.0.0 | 14 March 2018, 21:21:29 UTC |
9ddd520 | Dan Ringwalt | 14 March 2018, 17:49:40 UTC | Add OMR dependencies to kokoro presubmit. (#1106) | 14 March 2018, 17:49:40 UTC |
c3eda3d | Jacob Howcroft | 13 March 2018, 00:58:40 UTC | Coconet: Utility to export and load SavedModel (#1098) * Add Coconet utility for exportingto / loading from a SavedModel. * Some fixes to models/coconet/export_saved_model.py. | 13 March 2018, 00:58:40 UTC |
68898c8 | Curtis "Fjord" Hawthorne | 13 March 2018, 00:22:18 UTC | lint (#1105) | 13 March 2018, 00:22:18 UTC |
8c60d5c | Ian Simon | 13 March 2018, 00:16:15 UTC | add program & is_drum field to performances, and min/max MIDI program to constants (#1103) * Add program & is_drum fields to performances. * add docstring & tests * add min and max MIDI program to constants | 13 March 2018, 00:16:15 UTC |
f74a1d6 | Ian Simon | 13 March 2018, 00:15:19 UTC | make residual argument required in rnn_cell function (#1104) | 13 March 2018, 00:15:19 UTC |
e80c4be | Ian Simon | 12 March 2018, 23:47:15 UTC | add hparams for using residual connections in MusicVAE (#1099) | 12 March 2018, 23:47:15 UTC |
785c438 | Ian Simon | 12 March 2018, 17:23:16 UTC | temporarily restrict tornado version in kokoro (#1101) | 12 March 2018, 17:23:16 UTC |
45e7ab4 | Curtis "Fjord" Hawthorne | 07 March 2018, 22:01:30 UTC | properly separate train data from test (#1092) | 07 March 2018, 22:01:30 UTC |
2a8698b | Adam Roberts | 07 March 2018, 21:30:06 UTC | Fix MusicVAE sampling when n > batch_size. (#1093) * Fix MusicVAE sampling when n > batch_size. * respond to reviewer * lint | 07 March 2018, 21:30:06 UTC |
88efa67 | Curtis "Fjord" Hawthorne | 07 March 2018, 18:06:17 UTC | Make it clear we support Python 3 (#1091) * Make it clear we support Python 3 * Add minor version info | 07 March 2018, 18:06:17 UTC |
1c37eaa | Curtis "Fjord" Hawthorne | 07 March 2018, 00:29:00 UTC | Document ABC support (#1089) | 07 March 2018, 00:29:00 UTC |
dcf5019 | Curtis "Fjord" Hawthorne | 07 March 2018, 00:14:31 UTC | Add dev package install instructions for rtmidi (#1086) Fixes #1071. | 07 March 2018, 00:14:31 UTC |
fff592f | Adam Roberts | 02 March 2018, 01:25:54 UTC | Allow for null pitchClasses value (#1085) | 02 March 2018, 01:25:54 UTC |
e3ed650 | Jacob Howcroft | 02 March 2018, 00:46:12 UTC | Add docs to coconet_sample.Generator (#1045) * Add more detailed docs to coconet_sample.Generator * Fix wrong type of Generator wmodel * Update with requested changes from PR | 02 March 2018, 00:46:12 UTC |
0028d0b | Ian Simon | 01 March 2018, 22:10:42 UTC | fix MusicVAE chord-conditioning to not give hints about end token (#1084) | 01 March 2018, 22:10:42 UTC |
5fb5139 | Jacob Howcroft | 01 March 2018, 21:37:58 UTC | Add internal notebook visibility to Coconet (#1083) Continuing with merge because errors are unrelated to this PR and have been fixed by another PR that just hasn't been merged into this PR's branch. | 01 March 2018, 21:37:58 UTC |
7a4e44c | Adam Roberts | 01 March 2018, 18:17:29 UTC | Fix sort bug and add dist/musicvae.js to github. (#1082) * Fix sort bug * Add musicvae.js | 01 March 2018, 18:17:29 UTC |
e8176c8 | Curtis "Fjord" Hawthorne | 01 March 2018, 17:46:24 UTC | switch to number instead of number|Long (#1081) * v0.0.3 * switch to number instead of number|Long * v0.0.9 | 01 March 2018, 17:46:24 UTC |
6618c66 | Adam Roberts | 01 March 2018, 03:08:08 UTC | Switch to NoteSequence in musicvae.js (#1078) * Switch to NoteSequence in musicvae.js * Switch to NoteSequence in musicvae.js | 01 March 2018, 03:08:08 UTC |
9dc4985 | Curtis "Fjord" Hawthorne | 01 March 2018, 00:39:55 UTC | Switch to dcodeIO/protobuf.js, implement quantizeNoteSequence (#1074) * sequences and unit tests * add * test * test * full test * add functionality * fully functional * constants * fix build * trying out dcode * start switch to dcodeIO * switch to dcodeio * fix * fix * port remaining unit tests * fix * correctly handle custom error classes * create @magenta/protobuff, fix core build for npm export tensorflow.magenta.INoteSequence from protobuff * change sequences and test to reference @magenta/protobuff * fix tsc build and package.json pointing to correct files * remove dist folder from js/ * rename @magenta/protobuff to @magenta/protobuf * updates for separate protobuf module * v0.0.8 * fixes * prepublish * clean up proto package | 01 March 2018, 00:39:55 UTC |
9d6c717 | Ian Simon | 28 February 2018, 23:50:15 UTC | don't add chords past the end of a NoteSequence (#1080) | 28 February 2018, 23:50:15 UTC |
73928ed | Curtis "Fjord" Hawthorne | 28 February 2018, 23:22:04 UTC | bump version (#1079) | 28 February 2018, 23:22:04 UTC |
bad6882 | Adam Roberts | 28 February 2018, 20:20:58 UTC | Update @magenta/music-vae API with NoteSequence-like I/O. (#1073) * Add NoteSequence-like interface to MusicVAE-js * rename package to music_vae for consistency * rename package to music_vae for consistency * Add missing files. * Replace tabs. * Replace tabs. * Rename package to music-vae * Add documentation to model.ts * add javadoc | 28 February 2018, 20:20:58 UTC |
63ad504 | Ian Simon | 28 February 2018, 18:58:27 UTC | support control sequences in split and hierarchical decoders (#1075) | 28 February 2018, 18:58:27 UTC |
2e7bc85 | Adam Roberts | 27 February 2018, 04:45:54 UTC | Clean up the code, fix tidying to avoid memory leaks, and use async properly. (#1070) * Tidying up a bit. * Tidying up a bit. * more tidying * more tidying * Remove bundle.js * Moved init to separate line for clarity * Responding to reviewer comments. * Responding to reviewer comments. | 27 February 2018, 04:45:54 UTC |
91ab4ed | Adam Roberts | 26 February 2018, 08:17:36 UTC | Fixes to dl 0.5 update. Added more complete demo. (#1069) * Add initial version of MusicVAE deeplearn implementation. * Add readme * Add readme * Add readme * Incorporating Kyle's suggestions. * Add @magenta prefix to package * 0.5 * Add more useful demo page. * Fixes to dl 0.5 update. Added more complete demo. * fix indent * fix indent | 26 February 2018, 08:17:36 UTC |
7a601e1 | Torin Blankensmith | 26 February 2018, 08:06:59 UTC | Updated MusicVAE to deeplearn 0.5.0 (#1068) * updated to deeplearn 0.5.0 * updated demo to deeplearn 0.5 | 26 February 2018, 08:06:59 UTC |
909758c | Curtis "Fjord" Hawthorne | 23 February 2018, 21:53:53 UTC | Initial @magenta/core npm javascript package (#1065) | 23 February 2018, 21:53:53 UTC |