HEAD | 22b4495 | Updated DynetSharp for .NET 6.0 (#1667) * added implicit parameter to expression casting * added option for device specification for inputs functions like ones, zeros. * Added GPU (and device specification) support in csharp wrapper * Updated documentation for GPU support * Updated markdown * added input functions by dimensions instead of value * added LeCunUniform and NoamOpt - c++,python,sharp, fixed bugs in dynetsharp, adjusted transformer example to work with opennmt model * added mini-batching for dynet sharp! Also, fixed bugs, added another example, updated to work with new optimizers * Fixed NoamOpt learning-rate curve, verified result * Reverted not dynetsharp changes * Removed added HeInitializer and NoamOpt * Updated vcxProj to NetCore * Fixed shuffle | 08 November 2023, 12:40:01 UTC |
refs/heads/add-rounding | cc2ddc6 | Fixed namespace bug | 13 November 2018, 12:09:54 UTC |
refs/heads/appveyor-unittests | 2330d3b | Python unit testing seems to need to be in build\python directory Former-commit-id: a7808967cc11345c9f05f1cb4a268e27cf76f409 | 24 October 2017, 21:32:30 UTC |
refs/heads/autobatch-dropout | 6f24985 | Added autobatching for dropout | 06 April 2018, 15:43:36 UTC |
refs/heads/autobatch-lstm-transpose | 038faaf | Autobatching improvement for LSTM, addition for transpose | 26 March 2018, 15:03:54 UTC |
refs/heads/batched-matmul | d1d93dc | Fixed batched matmul | 22 April 2018, 02:57:29 UTC |
refs/heads/better-lstm-init-doc | 3b6c3b4 | Add documentation on the initialization of the VanillaLSTM's parameters | 13 September 2018, 15:14:35 UTC |
refs/heads/boundary-check-pick-batch-elem | 3cc21bb | Updated boundary check Former-commit-id: d86f5fa4ee74495790833b54a1a425304e2055b2 | 05 November 2017, 00:55:40 UTC |
refs/heads/broadcast-speed | acacaa6 | Fixed autobatch profile for csum | 09 December 2017, 00:43:51 UTC |
refs/heads/cuda-arch | da9a971 | Make possible to set CUDA compute architecture | 24 November 2017, 04:03:31 UTC |
refs/heads/curand | 68ce61d | Merge branch 'master' of github.com:clab/dynet into curand | 19 January 2018, 16:08:30 UTC |
refs/heads/debug-parallel-memcpy | f346cfe | Enabled print | 08 December 2017, 04:30:29 UTC |
refs/heads/dependabot/pip/examples/variational-autoencoder/basic-image-recon/numpy-1.22.0 | 1cec5ce | Bump numpy in /examples/variational-autoencoder/basic-image-recon Bumps [numpy](https://github.com/numpy/numpy) from 1.14.2 to 1.22.0. - [Release notes](https://github.com/numpy/numpy/releases) - [Changelog](https://github.com/numpy/numpy/blob/main/doc/HOWTO_RELEASE.rst) - [Commits](https://github.com/numpy/numpy/compare/v1.14.2...v1.22.0) --- updated-dependencies: - dependency-name: numpy dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> | 21 June 2022, 21:18:25 UTC |
refs/heads/dependabot/pip/examples/variational-autoencoder/basic-image-recon/pillow-10.0.1 | 9949bcd | Bump pillow in /examples/variational-autoencoder/basic-image-recon Bumps [pillow](https://github.com/python-pillow/Pillow) from 9.0.0 to 10.0.1. - [Release notes](https://github.com/python-pillow/Pillow/releases) - [Changelog](https://github.com/python-pillow/Pillow/blob/main/CHANGES.rst) - [Commits](https://github.com/python-pillow/Pillow/compare/9.0.0...10.0.1) --- updated-dependencies: - dependency-name: pillow dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> | 03 October 2023, 23:37:50 UTC |
refs/heads/disable-implicit-batching | bec6523 | Disable implicit batching | 29 October 2018, 20:11:44 UTC |
refs/heads/dump | 8eb5811 | cuda implementation of is_valid | 05 May 2019, 07:33:01 UTC |
refs/heads/dynamictreelstms | 01590ef | removed capitalization Former-commit-id: 4eee1eccf6a6fb98e0f0ac1aab26f4c16240d2a2 | 10 May 2016, 18:29:11 UTC |
refs/heads/empty-check | fa85459 | Check for empty lists in function calls | 16 February 2018, 16:42:02 UTC |
refs/heads/enable-win-io-test | 6cd33b0 | Re-enable test-io unit test on Windows after Chunyang's fix #1014. Former-commit-id: 8bf504196a13e3f24a3d91348ec1351c175b0d42 | 25 October 2017, 22:18:57 UTC |
refs/heads/fast-input-tensor | 2f9966f | fix negative indexing | 24 January 2018, 15:09:49 UTC |
refs/heads/faster-affine-bwd | 6c2e8a2 | Faster backward pass in affine trans w/ broadcasted bias | 25 October 2018, 16:04:40 UTC |
refs/heads/faster-bmm | 0d75f65 | Faster backward pass in affine trans w/ broadcasted bias | 25 October 2018, 15:57:32 UTC |
refs/heads/fix-appveyor | a50b512 | Fix appveyor boost version | 20 June 2018, 21:27:41 UTC |
refs/heads/fix-autobatch-corner | 538fd5c | Fixed corner case in autobatched exec | 25 February 2018, 14:24:21 UTC |
refs/heads/fix-autos | effb265 | Fixed errors on mac by removing "auto" calls | 17 January 2018, 14:58:11 UTC |
refs/heads/fix-block-dropout | bac9967 | Fix block dropout on GPU | 18 September 2018, 20:33:13 UTC |
refs/heads/fix-cuda-9.0 | b765813 | Fixes compile on CUDA 9.0 | 24 November 2017, 21:14:31 UTC |
refs/heads/fix-full-grad-populate | 20bfc35 | Fix bug in parameter loading with nonzero gradient | 14 September 2018, 18:59:19 UTC |
refs/heads/fix-highdim-lookup | 11b8aff | Fixed high-dimensional tensor lookup | 09 January 2019, 17:48:42 UTC |
refs/heads/fix-parallel-memcpy | 3a35c66 | Reverted to async and added stream | 08 December 2017, 16:23:16 UTC |
refs/heads/fix-rand-normal | 2a9bf9c | Fixed random normal and made mean/stddev an option | 05 February 2018, 15:03:34 UTC |
refs/heads/fix-stale-lookup | 67c5b9e | Fix stale lookup parameters | 20 June 2018, 21:08:36 UTC |
refs/heads/fix-travis-hg | f69a5fb | Remove mercurial call from travis to fix error | 10 May 2019, 14:47:16 UTC |
refs/heads/fix_loss_gradient | ff1fe41 | Fix broken thing | 23 February 2018, 15:43:00 UTC |
refs/heads/flexible-maxpooling2d | 3969220 | Fixed copying of dimension | 13 November 2017, 22:46:30 UTC |
refs/heads/force-inline | 62ce35a | Fix compile error on gcc 4.9 + CUDA 7.5 | 24 November 2017, 20:05:32 UTC |
refs/heads/full-multi-gpu-support | 6d19f4b | Address comments in pr Former-commit-id: 53216199c5861a260f9398c5587dba4b652099b0 | 15 October 2017, 16:37:22 UTC |
refs/heads/gitversion | 9edd9fc | add __gitversion__ info to python build | 07 February 2018, 19:43:17 UTC |
refs/heads/install-cudnn | 3caa301 | Added CuDNN to Python documentation Former-commit-id: e5d456d63bdfd5f21c1a45aa3483499550c2289f | 05 November 2017, 09:47:11 UTC |
refs/heads/joint_init | 5c3584c | Add joint initializer Former-commit-id: f61efe4932f12cde27abacab57562971de2225df | 06 March 2017, 18:21:20 UTC |
refs/heads/logsumexp-dim | 428cc07 | Fixed build on windows? Former-commit-id: acca1e13bd334b786bf5194d2806b70beda76c03 | 19 October 2017, 12:30:26 UTC |
refs/heads/master | 22b4495 | Updated DynetSharp for .NET 6.0 (#1667) * added implicit parameter to expression casting * added option for device specification for inputs functions like ones, zeros. * Added GPU (and device specification) support in csharp wrapper * Updated documentation for GPU support * Updated markdown * added input functions by dimensions instead of value * added LeCunUniform and NoamOpt - c++,python,sharp, fixed bugs in dynetsharp, adjusted transformer example to work with opennmt model * added mini-batching for dynet sharp! Also, fixed bugs, added another example, updated to work with new optimizers * Fixed NoamOpt learning-rate curve, verified result * Reverted not dynetsharp changes * Removed added HeInitializer and NoamOpt * Updated vcxProj to NetCore * Fixed shuffle | 08 November 2023, 12:40:01 UTC |
refs/heads/matmul_mult_sigs | 06ec749 | introduce a version of matmul which does not depend on first arg Former-commit-id: 46c9a58f8b8ee3650bccdb31424a99f61994bdd6 | 25 May 2017, 23:08:20 UTC |
refs/heads/move-eigen-header | 88853d6 | Merge branch 'master' of github.com:clab/dynet into move-eigen-header | 19 December 2017, 17:51:57 UTC |
refs/heads/multithread | 3be5f4a | multi-thread working, no speedup on sequence benchmark Former-commit-id: 71acd1a262883a57513176a826fbb5cc52ffd2b2 | 27 May 2017, 13:30:40 UTC |
refs/heads/param_expr_update_flag | a219de2 | Clarify documentation | 18 April 2018, 17:34:54 UTC |
refs/heads/param_expression | 2090525 | Fixed deprecation message. | 01 March 2018, 15:00:53 UTC |
refs/heads/pickneglogsoftmax-boundary-check | 94b3f34 | Bug fix Former-commit-id: 9d67614cb27b3270962e8630156df969b1da6678 | 26 October 2017, 13:53:38 UTC |
refs/heads/profiling | b452c3b | fix previous commit | 23 November 2017, 12:15:47 UTC |
refs/heads/profiling_arg | 0729050 | --dynet-profiling was not removing its argument. | 27 November 2017, 23:22:34 UTC |
refs/heads/protobuf | 4c25eb0 | Merge pull request #150 from mattr1/mattri/f1_assert Change "assert(false)" to "throw exception(...)" so it always fails, and also compiles properly with NDEBUG set Former-commit-id: 82ad93744f37daa35a5dda4dec6cb918be8fe1bb | 08 November 2016, 16:49:54 UTC |
refs/heads/python-doc-upgrade | d99742e | Added mention of binary distribution | 09 February 2018, 21:43:06 UTC |
refs/heads/python_weight_decay_access | 9fc219a | weight decay inherited from parent model | 12 February 2018, 16:23:01 UTC |
refs/heads/random-gpu | 01f4548 | Merge branch 'master' into random-gpu | 19 December 2017, 15:33:51 UTC |
refs/heads/range-improvements | 4276b35 | fix test name | 11 May 2018, 12:05:15 UTC |
refs/heads/read_inplace0 | 7b6005a | Removed extra message | 10 January 2018, 17:29:43 UTC |
refs/heads/reset_random_seed | 43f7b5a | First commit | 17 April 2018, 19:46:28 UTC |
refs/heads/revert-eigen-stream | 7a89155 | Fixed compile error | 25 November 2017, 03:55:55 UTC |
refs/heads/rnn_states | b75c252 | fix initializing RNN state with given expressions | 07 February 2018, 20:57:36 UTC |
refs/heads/sequence-ops | ce60e4b | Merge branch 'master' of https://github.com/clab/dynet into sequence-ops Former-commit-id: 1ccb55e7e98d37d2d76a5edc44710f9dcf97572a | 04 January 2017, 14:02:17 UTC |
refs/heads/set-weight-decay-lambda | 26f2043 | update doc | 08 January 2018, 14:59:30 UTC |
refs/heads/stateful-trainer-multidevice | fe115bd | Fixed stateful trainers to work with multi-device | 24 November 2017, 16:44:39 UTC |
refs/heads/strided-select2 | b67702c | fix compiling on windows | 14 November 2017, 16:56:13 UTC |
refs/heads/sum-dim | 8bea6de | Fixed default batch value Former-commit-id: 8b480d7fc72afdc6ea7298f840e80e3a75c8af25 | 18 October 2017, 21:36:58 UTC |
refs/heads/test-doc | 1d7b45c | examples: attempt to fix training with standard softmax (#905) BuildLMGraph assumes "cfsm" points to an object of class factored softmax, but it can point to an object of StandardSoftmaxBuilder if users run the command as documented in README.md. I'm not sure this is a correct fix because the comments in BuildLMGraph use a bit different term, "regular softmax". If it refers to the standard softmax, removing allocating StandardSoftmaxBuilder object seems correct. Former-commit-id: 53c47b863fe27262a9795dad7588e64250901b3f | 22 September 2017, 12:21:26 UTC |
refs/heads/upgrade-eigen-instructions | c09cb9a | Upgraded Eigen version to fix CUDA9.1 compile | 16 April 2018, 20:16:47 UTC |
refs/heads/v1 | d111837 | Updated dynet to use initialization Former-commit-id: b0041d78bd593f7956cefed0bf41d1fecc402334 | 11 October 2016, 23:21:28 UTC |
refs/heads/v1.1 | 97b9276 | Merge pull request #650 from danielhers/v1.1 Update version in setup.py Former-commit-id: 2d779f9f22f4754726561d31e4cd43486c2a8933 | 03 July 2017, 19:50:52 UTC |
refs/heads/weight_decay_at_params | 802ab16 | weight decay pushed to parameters level (cpp) | 05 March 2018, 11:05:03 UTC |
refs/reviewable/pr851/r1 | 695d363 | dynet-849. Fix GRUBuilder self._spec bug. Former-commit-id: bd971efa634c978f29e76c890b195c00c44c624e | 02 September 2017, 14:39:11 UTC |
refs/tags/2.0.1 | 59b77f7 | dynet-849. Fix GRUBuilder self._spec bug. (#851) Former-commit-id: 87df34103625102493f8c660684146a636e2482c | 02 September 2017, 14:49:51 UTC |
refs/tags/2.0.2 | 9b50b35 | underflow problems were causing some spurious failures with the new stricter grad check | 21 December 2017, 05:10:52 UTC |
refs/tags/2.0.3 | 7cb6455 | Added mention of binary distribution (#1234) | 09 February 2018, 21:54:04 UTC |
refs/tags/2.1 | d9d3e80 | Fix glorot initialization for convolutional kernels (#1420) * fix glorot initialization for convs * Update Docs * Updated benchmark results | 18 September 2018, 14:51:16 UTC |
refs/tags/2.1.1 | 7be57e6 | Upload to PyPI with long description in Markdown, not RST | 21 October 2020, 11:02:41 UTC |
refs/tags/2.1.2 | b564545 | Remove pandoc dependency, not needed since PyPI supports Markdown | 21 October 2020, 13:33:17 UTC |
refs/tags/v1.0-rc1 | d111837 | Updated dynet to use initialization Former-commit-id: b0041d78bd593f7956cefed0bf41d1fecc402334 | 11 October 2016, 23:21:28 UTC |
refs/tags/v1.1 | 363a8cc | Small fix to doc Former-commit-id: 78f7bedf74f530586f2993d0c3b96b0b6606e590 | 28 June 2017, 01:25:20 UTC |
refs/tags/v2.0 | 1241cfc | Merge pull request #679 from clab/nodes-separate-final Separate nodes into different files Former-commit-id: 9606e2a3fdc31788b3739adfcfb498b1626645fd | 10 July 2017, 16:38:55 UTC |
refs/tags/v2.1.2 | b564545 | Remove pandoc dependency, not needed since PyPI supports Markdown | 21 October 2020, 13:33:17 UTC |