https://github.com/clab/dynet

sort by:
Revision Author Date Message Commit Date
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
c418b09 Travis and Appveyor build for Python 3.9 and 3.10 14 August 2022, 17:11:42 UTC
3e1b48c fixes executable name in readme set up instructions (#1572) 11 March 2022, 14:08:21 UTC
63725e9 The following error is often thrown out when multiple users run multi-process training on a machine (across the time), which happens a lot in a cluster environment. (#1633) terminate called after throwing an instance of 'boost::interprocess::interprocess_exception' what(): Permission denied What happened is: User A run a multi-process training. A queue file with name "dynet_mp_work_queue" is created with each multi-process training, and is often not deleted properly due to manual termination (like CTRL+C). Now User B tries to run a multi-process training on the same machine, and could not write/delete the queue file. So the above error is thrown. Fix: Include process id in the queue file name, so different trainings by different users do not step on each other's toes. 11 March 2022, 14:07:50 UTC
6658c73 Bump pillow in /examples/variational-autoencoder/basic-image-recon (#1645) Bumps [pillow](https://github.com/python-pillow/Pillow) from 7.1.0 to 9.0.0. - [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/7.1.0...9.0.0) --- updated-dependencies: - dependency-name: pillow dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> 11 March 2022, 14:07:13 UTC
a792752 Avoid -march=native on Macs with M1 (#1648) This flag causes issues on M1 Macs, as clang does not support this flag on this platform. Instead, use -march-apple-m1. 11 March 2022, 14:07:00 UTC
93a5cd2 Bump pillow in /examples/variational-autoencoder/basic-image-recon (#1631) Bumps [pillow](https://github.com/python-pillow/Pillow) from 6.2.0 to 7.1.0. - [Release notes](https://github.com/python-pillow/Pillow/releases) - [Changelog](https://github.com/python-pillow/Pillow/blob/master/CHANGES.rst) - [Commits](https://github.com/python-pillow/Pillow/compare/6.2.0...7.1.0) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> 27 January 2021, 15:43:03 UTC
04ffe21 Make sure ParameterStorageBase also uses shared pointer for Java interface (#1576) Its subclasses are marked in dynet_swig.i as using shared pointers: ``` %shared_ptr(dynet::ParameterStorage) %shared_ptr(dynet::LookupParameterStorage) ``` SWIG warns that the base class "is not similarly marked as a smart pointer". The line below quiets the warning and also prevents crahes (because of double deletes?). ``` %shared_ptr(dynet::ParameterStorageBase) ``` An issue was filed previously and this will soon link to it. 27 January 2021, 15:42:30 UTC
b6565c6 Protect Scala Expression from being deleted via its ExpressionVector (#1629) If an ExpressionVector is garbage collected, the C++ memory for the vector and its expressions will be freed in C++. If any Scala Expressions that were derived from it with apply(n) are still alive, their C++ counterparts will have disappeared out from under them. The Scala Expression, when used, will try to access its C++ counterpart and cause the C++ code to crash and take down the JVM with it. 27 January 2021, 15:42:08 UTC
91fdb95 Fix some typos in documentation (#1630) 27 January 2021, 15:41:20 UTC
480d95a Free C++ memory associated with Scala Pointers (#1628) 27 January 2021, 15:40:51 UTC
ef1190d Add indices checking for parameters in model.cc (#1593) Currently, there is no indices checking in gradient_l2_norm_dev and sometimes the code could request elements beyond the end of the vector (when k1 = params.size() or k2 = lookup_params.size()). To fix this, we introduce a simple bounds check. params.size() > k1 lookup_params.size() > k2 17 January 2021, 19:32:57 UTC
13eb330 Free memory for the correct variable (#1627) Otherwise, memory of project_scratch is leaking and gradient_norm_scratch use will cause problems. 17 January 2021, 19:32:36 UTC
0781dea Update Travis dist for Python 3.7 and 3.8 pip tests 22 October 2020, 07:16:12 UTC
cad7618 Remove Python 3.4 tests, no supported 22 October 2020, 07:14:19 UTC
b564545 Remove pandoc dependency, not needed since PyPI supports Markdown 21 October 2020, 13:33:17 UTC
7be57e6 Upload to PyPI with long description in Markdown, not RST 21 October 2020, 11:02:41 UTC
5ce3a39 Check wheel before deploying it 21 October 2020, 10:03:20 UTC
b2f8509 Fix eigen unpacking directory 21 October 2020, 10:02:53 UTC
e28cd90 Verbose output when deploying 20 October 2020, 14:19:38 UTC
8e1fdc3 Remove Python 3.5 from AppVeyor 20 October 2020, 09:32:21 UTC
3c76552 Support Python 3.8 06 October 2020, 10:08:42 UTC
9fed650 Get Eigen zip from GitHub release 06 October 2020, 10:03:35 UTC
4e28dc8 Fix Eigen commit 24 August 2020, 12:35:37 UTC
4225e62 Fix Eigen filename 24 August 2020, 11:38:03 UTC
b800ed0 Eigen moved to GitLab 24 August 2020, 11:11:16 UTC
32c71ac Install cryptography from binary on Travis docker too 18 August 2020, 11:05:41 UTC
0fc7907 Upgrade pip on Travis to allow --prefer-binary 18 August 2020, 11:01:49 UTC
7c533e5 Fix build error on Travis CI test for i686 pyca/cryptography#5069 31 July 2020, 13:06:39 UTC
c61bc99 Remove Python 3.4 Travis CI tests due to Pygments version conflict 31 July 2020, 13:01:32 UTC
0679789 Build for Python 3.8 on AppVeyor 07 July 2020, 13:56:24 UTC
85a4664 Build for Python 3.8 on Travis CI 07 July 2020, 13:55:00 UTC
352653b Fix serious memory leak in code using CUDNN functionality (#1614) * More useful Debug and Release targets. Added ReleaseWithDebugInfo target which is needed to find memory leaks. * Support for CUDNN V8 * Memory leaks fixed. 07 July 2020, 01:33:37 UTC
1a5e311 Add a blurb about fatdynet as suggested (#1602) Feel free to edit. Thanks. 04 March 2020, 19:48:11 UTC
f197aac Add PEP 518 build system/requirements spec (#1608) 04 March 2020, 19:47:49 UTC
fcf3e4c Fix openssl error in Docker builds too 02 March 2020, 10:10:47 UTC
99b7d6a Fix openssl error in Docker builds too 14 February 2020, 12:33:26 UTC
4e893ee Remove Python 3.4 Travis tests on Mac, as 3.5 or later is required 14 February 2020, 12:26:15 UTC
5573434 Fix fatal error: openssl/opensslv.h: No such file or directory https://stackoverflow.com/a/46008642/223267 17 January 2020, 12:57:17 UTC
7dfa70e Use conda-forge to support Python 3.4 https://stackoverflow.com/a/57113184/223267 10 January 2020, 13:37:02 UTC
b11ce5c Fix build failure pypa/manylinux#357 10 January 2020, 13:29:39 UTC
fa5d7f0 cblas sgemm batch (#1583) 29 October 2019, 13:52:46 UTC
9966917 Bump pillow in /examples/variational-autoencoder/basic-image-recon (#1596) Bumps [pillow](https://github.com/python-pillow/Pillow) from 5.0.0 to 6.2.0. - [Release notes](https://github.com/python-pillow/Pillow/releases) - [Changelog](https://github.com/python-pillow/Pillow/blob/master/CHANGES.rst) - [Commits](https://github.com/python-pillow/Pillow/compare/5.0.0...6.2.0) Signed-off-by: dependabot[bot] <support@github.com> 29 October 2019, 13:52:22 UTC
c64c86f Let the user set CUDA_TOOLKIT_ROOT_DIR on install When running BACKEND=cuda pip install git+https://github.com/clab/dynet#egg=dynet If CUDA is installed in a non-standard location, cmake will complain: INFO:root:Configuring... -- Optimization level: fast -- BACKEND: cuda CMake Error at /usr/share/cmake/Modules/FindCUDA.cmake:548 (message): Specify CUDA_TOOLKIT_ROOT_DIR Call Stack (most recent call first): CMakeLists.txt:160 (find_package) Instead of having to perform manual installation, this will allow setting CUDA_TOOLKIT_ROOT_DIR by an environment variable. 14 August 2019, 08:01:24 UTC
a581587 [update] In order to avoid misunderstanding regarding to prediction function, a argument of forward() was changed. (#1568) 19 June 2019, 00:25:19 UTC
72211da Prediction is added to train_xor.cc (#1565) * [update] prediction is added to train_xor.cc * Comment is changed 14 June 2019, 16:55:57 UTC
32fc1b6 [typo-fix] In expr.h, tuntime_error -> runtime_error (#1564) 06 June 2019, 15:52:00 UTC
1ae19fd Fix sign in description of log_sigmoid (#1563) Documentations appears wrong; this fixes the sign in the description of `log_sigmoid` 06 June 2019, 15:51:29 UTC
1fafb23 GPU support for C# + added device specification for numeric ecpression functions. (#1553) * 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 29 May 2019, 12:24:55 UTC
7271b81 Add Rust bindings (#1557) * Implement Dim for Rust-bindings. * Implement DynetParams for Rust-bindings. * Implement Device for Rust-bindings. * Implement ParameterInit for Rust-bindings. * Typo. * Implement Tensor for Rust-bindings. * Implement Parameter, LookupParameter, ParameterCollection for Rust-bindings. * Fix styles. * Implement Trainer for Rust-bindings. * Implement Expression for Rust-bindings. * Implement ComputationGraph for Rust-bindings. * Fix Trainer. * Implement input operations for Rust-bindings. * Implement arithmetic operations for Rust-bindings. * Fix C-APIs. * Implement probability/loss operations for Rust-bindings. * Implement flow/shaping operations for Rust-bindings. * Implement noise operations for Rust-bindings. * Fix operation. * Implement tensor, linear algebra, convolution, normalization, and device operations for Rust-bindings. * Implement io for Rust-bindings. * Fix documents. * Add xor example. * Fix dim. * Implement clone for dynetExpression. * Implement Clone for Expression in Rust-bindings. * Implement MNIST example for Rust-bindings. * Fix warnings for Rust-bindings. * Implement RNNBuilder for Rust-bindings. * Implement SoftmaxBuilder for Rust-bindings. * Implement RNNLM example for Rust-bindings. * Implement expr functions for C-APIs. * Implement ComputationGraph::dump for C-APIs. * Implement APIs for Rust-bindings. * Add docs for Rust-bindings. * Add lints for Rust-bindings. * Update build script for Rust-bindings. * Fix build script for Rust-bindings. * Fix build script for Rust-bindings. * Fix RNNBuilder for Rust-bindings. * Add features. * Update documents. 29 May 2019, 00:26:26 UTC
95e9dd9 Make sure shared library looks like one (#1551) Windows (10) will not do a System.load on a file that is not named like a loadable file. 18 May 2019, 02:41:58 UTC
301ecd3 Prevent overflow in calculation of strsize for [lookup]parameters (#1550) The integer calculation otherwise overflows before conversion to a size_t, resulting in bad values being written to serialized models. 17 May 2019, 13:31:29 UTC
b9c5f65 Remove mercurial call from travis to fix error (#1545) 10 May 2019, 16:29:10 UTC
8ee8bf6 Dump (#1542) * Added function to dump the current values and/or gradients of nodes in the computation graph either to stdout or to a file. * cuda implementation of is_valid 10 May 2019, 14:36:41 UTC
e934e92 A bracket was not closed in cpp_basic_tutorial.rst. (#1544) from sigmoid function logistic regression). to sigmoid function (logistic regression). On branch document_unclose_bracket Changes to be committed: modified: cpp_basic_tutorial.rst 08 May 2019, 13:05:47 UTC
62718ad moved to keeping variable-index instead of storing expression object, fixed AccessViolation bug when inputting external values (#1543) 05 May 2019, 23:48:03 UTC
640b13d Added function to dump the current values and/or gradients of nodes in the computation graph either to stdout or to a file. (#1541) 03 May 2019, 12:50:03 UTC
c7abce3 Initial commit for dynet wrapper for c# - dynetsharp (#1540) * initial commit for dynet wrapper for c# - dynetsharp * added examples, readme.md * updated readme.md * changed all uints to ints * fixed rnnlm example * updated readme * fixed fRecalculate flag with parameter objects 01 May 2019, 17:55:14 UTC
21cc626 fix current version (#1506) (#1523) 30 April 2019, 00:49:13 UTC
4a7fcd9 Update install.rst (#1526) Change ./examples/train_xor to ./examples/xor. Running ./examples/train_xor produces an error "No such file or directory" 30 April 2019, 00:48:56 UTC
af83e46 Removed redundant comment on include safety (#1524) Removed redundant comment on include safety due to header guards 30 April 2019, 00:48:38 UTC
2c58ad9 Fix dynet.conv2d_bias doc on the shape of b (#1531) 30 April 2019, 00:48:19 UTC
e5a1116 copy implementation of logistic sigmoid from Eigen 08 March 2019, 17:50:23 UTC
28c133a Revert "use Eigen's logistic sigmoid functor instead of broken one" This reverts commit c125ecde3e873c98ac22f1adc19359510a8fdc82. 08 March 2019, 17:21:10 UTC
28eaf94 Revert "use more recent eigen" This reverts commit 096956560280a1ae78037a0337e209302513515e. 08 March 2019, 17:20:58 UTC
539ab76 Revert "use new device types with new eigen" This reverts commit a17c72e7289afaa1fff2dd5ae484beff9cdc9354. 08 March 2019, 17:20:46 UTC
a17c72e use new device types with new eigen 08 March 2019, 14:45:54 UTC
0969565 use more recent eigen 08 March 2019, 14:45:32 UTC
c125ecd use Eigen's logistic sigmoid functor instead of broken one 08 March 2019, 14:00:51 UTC
bc1e70e fix ambiguous indentation 08 March 2019, 12:06:34 UTC
d65bd5e Fixed high-dimensional tensor lookup (#1514) 09 January 2019, 18:34:44 UTC
946d440 Reinforcement Learning using DyNet: DQN and DDPG (#1394) * lower-case repo name * change the height of 1 gif * revise readme * Update README.md * update broken links to result gifs * Update README.md * Ore-organize code and fix some code quality problem. * remove ./idea 08 January 2019, 22:22:42 UTC
53b0464 Exponential Moving Average (#1466) * First EMA prototype * update w. master * IO adaptation for EMA + tests * EMA example * EMA example * DYNER_ASSERT -> DYNET_RUNTIME_ERR * fix bugs reported by alexandres * Implement both exponential and cumulative moving average * change saving strategy * EMA/CMA example 08 January 2019, 22:22:09 UTC
19b8622 Disable implicit batching (#1493) 08 January 2019, 22:20:42 UTC
2e42045 add note about dedault gradient clipping behavior (#1511) 08 January 2019, 22:20:11 UTC
8b334ce add __host__ qualifier to functors 20 December 2018, 14:37:47 UTC
51b528c cmake option for custom opt level (#1497) Thanks! 13 November 2018, 15:01:42 UTC
564538a Added round, ceil, and floor functions (#1494) * Added round, ceil, and floor functions * Added missing files * Removed extraneous includes * Fixed namespace bug 13 November 2018, 15:01:06 UTC
8260090 run_single_process dev loss is accumulated, run_multi_process no cleanup (#1415) * run_single_process dev loss is accumulated across dev batches #1414 * resource leak in run_multi_process - pipes not closed #1419 29 October 2018, 17:58:16 UTC
c86b484 fixed softmax latex typos (#1487) 29 October 2018, 17:55:58 UTC
946200c Matmul code and speed improvements (#1483) * Initial commit: remove code duplication * Fix transposition order bug * Remove the mysterious assert 25 October 2018, 13:28:49 UTC
bc3ff25 Merge pull request #1476 from ruyimarone/master fixed tanh escape typos 15 October 2018, 14:15:29 UTC
b691dd6 fixed tanh escape typos 12 October 2018, 07:00:55 UTC
d7ff185 Merge pull request #1472 from chantera/fix/default-device Throws exception when `default_device` is null. 08 October 2018, 19:41:27 UTC
9a0a4f3 Merge pull request #1468 from chantera/fix/docs Fix link in documentation. 08 October 2018, 18:42:55 UTC
8d51896 Throws exception when `default_device` is null. 08 October 2018, 18:32:54 UTC
f2f13e7 Fix link in documentation. 07 October 2018, 01:47:41 UTC
ab6ea34 Merge pull request #1464 from bskaggs/docs Make capitalization of "DyNet" consistent in documentation. 05 October 2018, 15:55:56 UTC
6040f5f Make capitalization of "DyNet" consistent in documentation. 05 October 2018, 03:16:02 UTC
af88f17 Merge pull request #1462 from chantera/feature/c-api Add C-APIs. 03 October 2018, 14:44:11 UTC
4e23948 Merge pull request #1459 from FilippoC/trainer-save-load Save/load the optimizer state 03 October 2018, 14:43:45 UTC
05c8856 Fix dimension vector type for dynetDim in C-API. 02 October 2018, 10:39:07 UTC
0d61453 Remove useless method+attribute in the trainers-io test 02 October 2018, 09:33:24 UTC
e72374c replace BOOST_TEST with BOOST_CHECK for Travis 02 October 2018, 09:30:49 UTC
33638c7 Fix travis build command. 02 October 2018, 08:28:42 UTC
b8df602 Add documentation. 02 October 2018, 08:10:50 UTC
118127f Add Readme. 02 October 2018, 08:02:11 UTC
de1e3ca fix for continuous integration 01 October 2018, 19:26:05 UTC
58242e8 Remove g++-7 ref in CMakeLists.txt 01 October 2018, 16:56:00 UTC
c2e6da3 Unit tests for optimizer io 01 October 2018, 16:54:23 UTC
back to top