Commit Graph

  • 344f3fdaab fix typo cfujitsang 2019-11-01 12:59:19 -07:00
  • cec1ab2790 remove bwd_ignore_zero_init in test_operator_gpu.py cfujitsang 2019-11-01 12:42:51 -07:00
  • 95cb2fd86f remove bwd_ignore_zero_init flag cfujitsang 2019-11-01 12:34:29 -07:00
  • a93ad94162 remove std::move for clang cfujitsang 2019-11-01 11:48:56 -07:00
  • 68ded77587 add error return if bwd_ignore_zero_init is used without MXNET_EXEC_ENABLE_ADDTO cfujitsang 2019-11-01 09:18:33 -07:00
  • f8bd1cb65f add warning in bwd_ignore_zero_init description, also test with fp32 cfujitsang 2019-10-31 07:33:42 -07:00
  • 0b25025743 add _contrib for namespace and add GPU only on description cfujitsang 2019-10-21 14:00:40 -07:00
  • e76c38a59c remove a wrong copy-paste cfujitsang 2019-10-21 11:59:39 -07:00
  • e88fa4b68c add equivalent mxnet code in description of mha ops cfujitsang 2019-10-21 11:11:31 -07:00
  • d4ffa4e824 only use cublasStridedBatchedGemm if CUDA >= 9.1 cfujitsang 2019-10-18 14:19:53 -07:00
  • 7f5194dfd9 remove tests using kAddTo cfujitsang 2019-10-18 11:26:16 -07:00
  • 0485ed247a Add all the cuda/cublas helper functions cfujitsang 2019-10-18 08:17:19 -07:00
  • 67f5aa934e correct another typo cfujitsang 2019-10-16 15:55:59 -07:00
  • 3115fb1bef fix typos cfujitsang 2019-10-16 09:44:43 -07:00
  • 7ef32eeab9 change credit for the approach since the code have changed cfujitsang 2019-10-15 17:29:17 -07:00
  • 9b5ee9a7ad add better error message and description and remove cutlass from compilation flags cfujitsang 2019-10-15 13:13:45 -07:00
  • 543098b9d6 remove all cutlass stuff cfujitsang 2019-10-15 10:07:26 -07:00
  • 6728de81e8 add cutlass to compilation flags cfujitsang 2019-10-09 12:52:32 -07:00
  • 77beeb6ffc add cutlass as 3rdparty dependency cfujitsang 2019-10-09 11:49:11 -07:00
  • c0560fc1e8 add log message and TODO Lin Yuan 2019-11-05 22:38:53 -08:00
  • bfea5094ed Embedding gradient performance optimization on GPU (#16355) MoisesHer 2019-10-05 15:59:36 -07:00
  • 82ed82f25d Aggregated zero grad (#16446) Andrei Ivanov 2019-10-15 09:06:19 -07:00
  • 8c22fac8f5 Aggregated adamw update (#16398) Andrei Ivanov 2019-10-19 16:51:23 -07:00
  • 0415a2f0c7 Eliminate common expressions (#15657) Przemyslaw Tredak 2019-11-01 09:51:19 -07:00
  • 51c2065c36 Pointwise fusion for GPU (#15167) Przemyslaw Tredak 2019-10-31 21:07:24 -07:00
  • 3c404a5128 Mixed data type binary ops (#16699) Hao Jin 2019-11-05 08:59:05 -08:00
  • 0c5677ed33 Faster GPU NMS operator (#16542) Przemyslaw Tredak 2019-11-04 23:38:50 -08:00
  • c583e44816 fix requantize flaky test (#16709) Xinyu Chen 2019-11-05 13:57:06 +08:00
  • b9f3b0618b Updated logos. (#16719) Thom Lane 2019-11-04 14:47:15 -08:00
  • 5a2fce5dff [WIP][New Op] Add deformable conv v2 (#16341) Joshua Z. Zhang 2019-11-04 14:33:39 -08:00
  • bb6305d11d [MKLDNN] support mkldnn gelu (#16710) Xinyu Chen 2019-11-04 18:11:09 +08:00
  • 94aab3948e [Quantization] Enhance gluon quantization API (#16695) Wuxun Zhang 2019-11-04 13:31:49 +08:00
  • e139442461 support mixed-precision true_divide (#16711) Hao Jin 2019-11-02 22:15:55 -07:00
  • 27a8fd5ab8 MKLDNN FC : Add error info when mkldnn fc bias dimension is wrong (#16692) rongzha1 2019-11-03 08:43:07 +08:00
  • be92cc63ce change the wording and log level to be more in line with the general use (#16626) Carin Meier 2019-11-02 14:29:06 -04:00
  • 7fde59a029 Fix formatting and typos in CD README.md (#16703) Kellen Sunderland 2019-11-02 07:17:26 -07:00
  • 904ba00b48 improve example docs with installation instructions (#16625) Carin Meier 2019-11-02 08:39:42 -04:00
  • 1210c7dd83 Fallback to official NumPy operators (#16698) reminisce 2019-11-01 15:25:21 -07:00
  • 0cbee04b65 Fix the index_t with int comparisoon Przemek Tredak 2019-10-25 09:48:26 -07:00
  • 2efe4a51a0 Fix for the mixed naming convention Przemek Tredak 2019-10-24 16:52:15 -07:00
  • 761663af43 Fix from rebase Przemek Tredak 2019-10-24 16:50:02 -07:00
  • f59a3e739d Fix sort Przemek Tredak 2019-10-24 16:23:52 -07:00
  • e3bdb2eb6c Fixes Przemek Tredak 2019-10-18 22:02:42 -07:00
  • 6aad918a4d Early exit in Triangle kernel Przemek Tredak 2019-10-18 21:19:14 -07:00
  • b85dbd419b Fix typo Przemek Tredak 2019-10-18 21:58:40 -07:00
  • a496fa17f3 Fix lint and more fixes from rebase Przemek Tredak 2019-10-18 21:10:35 -07:00
  • 70b6e4d86e Fixes from rebase Przemek Tredak 2019-10-18 20:51:47 -07:00
  • b754289646 Cleaning Przemek Tredak 2019-10-18 15:29:40 -07:00
  • 85f1c319f9 Sort only topk results during second sorting Przemek Tredak 2019-10-18 14:55:11 -07:00
  • a90446282f Do not recompute other boxes area every time Przemek Tredak 2019-10-18 13:46:10 -07:00
  • 581e119220 Optimizations pt3 Przemek Tredak 2019-10-18 13:28:18 -07:00
  • 32144edaa6 Optimization pt2 Przemek Tredak 2019-10-17 15:36:44 -07:00
  • 955552ea87 Adding out-of-place ability to SortByKey Przemek Tredak 2019-10-17 14:45:46 -07:00
  • 9e14fe655a Optimization Przemek Tredak 2019-10-17 13:11:16 -07:00
  • 15b53fae80 Removing second sort Przemek Tredak 2019-10-17 11:10:40 -07:00
  • fec58ae93c NMS kernel Przemek Tredak 2019-10-14 15:41:23 -07:00
  • 6feb3b6d5b Fixed some formatting and typos in CD Readme. KellenSunderland-patch-1 Kellen Sunderland 2019-11-01 12:49:10 -07:00
  • aa491f2f9f [NumPy] Add NumPy support for Cholesky decomposition (#16667) vexilligera 2019-11-01 17:43:36 +00:00
  • 1aa1b5a9ab Eliminate common expressions (#15657) Przemyslaw Tredak 2019-11-01 09:51:19 -07:00
  • 9f6070f560 [Estimator] refactor estimator and clarify docs (#16694) Sheng Zha 2019-11-01 08:57:32 -07:00
  • f3c6be594e Revert "[MKLDNN]Fix reorder2default (#16602)" (#16697) Zhennan Qin 2019-11-01 20:58:02 +08:00
  • c01375741f [MKL-DNN] Add mxnet mkldnn cmake tutorial (#16688) Xinyu Chen 2019-11-01 20:56:36 +08:00
  • b5c2b82c17 fix binary dependencies in CD and nightly (#16693) Tao Lv 2019-11-01 20:54:51 +08:00
  • 33d108ba02 Initializer.__eq__ (#16680) Leonard Lausen 2019-11-01 02:37:10 -07:00
  • 6c42992f2a [numpy] add numpy operator : append (#16564) JiangZhaoh 2019-11-01 16:20:06 +08:00
  • 9f6e819d38 fix install dir (#16690) Tao Lv 2019-11-01 14:36:13 +08:00
  • 4149f8b875 Pointwise fusion for GPU (#15167) Przemyslaw Tredak 2019-10-31 21:07:24 -07:00
  • ef19b09c29 [Estimator] refactor estimator to allow overriding evaluate/fit of a batch (#16678) Sheng Zha 2019-10-31 15:45:39 -07:00
  • 9cb896014e [Estimator] handle composite metrics in estimator (#16676) Sheng Zha 2019-10-31 15:26:00 -07:00
  • 979e610281 Added launch bounds to the reduce kernels (#16397) Przemyslaw Tredak 2019-10-31 11:10:06 -07:00
  • aa1074dc17 Upgrade MKL-DNN dependency to v1.0 (#16555) Tao Lv 2019-10-31 22:55:13 +08:00
  • b5d07e3032 Add check if scipy is imported in sparse.py (#16574) Leonard Lausen 2019-10-31 13:56:43 +08:00
  • f9baec9a02 [Numpy] implement np.column_stack (#16594) vexilligera 2019-10-31 04:59:27 +00:00
  • a6a9706bf9 Miscellaneous fix for several numpy issues (#16664) Hao Jin 2019-10-30 21:02:06 -07:00
  • 77e8f516e7 fix cuDNN RNN dtype_with_fallback_ bug (#16671) Jake Lee 2019-10-30 16:33:15 -07:00
  • 27bddf8f16 [Numpy] Numpy operator diff (#15906) Zhiqiang Xie 2019-10-30 17:34:50 +08:00
  • 8e50fd9d86 no such method => modified function args (#16610) phinzphinz 2019-10-30 01:55:02 +01:00
  • 5aa74e0662 Move ops which don't support FP16 dtype to FP32 list (#16668) Anirudh Subramanian 2019-10-29 13:30:57 -07:00
  • 60d74bc948 Showing proper error message when an attempt is made to create large tensor but MXNet is not built with it (#16570) Rohit Kumar Srivastava 2019-10-28 21:59:58 -07:00
  • 86ed5f5c0a [NumPy][Operator] NumPy operator may_share_memory and shares_memory (#16533) Huang, Guangtai 2019-10-28 16:24:05 +08:00
  • 9322864934 Infra for tvm op runtime dispatch (#16100) Haozheng Fan 2019-10-28 14:02:49 +08:00
  • 11dff51f71 Fix GetMKLDNNData for delay alloc (#16618) Zhennan Qin 2019-10-28 10:11:28 +08:00
  • ffc5392d6c Disable float16 test (#16643) Haozheng Fan 2019-10-28 07:36:26 +08:00
  • 6ab4220e8d numpy doc enhancement (#16637) reminisce 2019-10-27 16:31:39 -07:00
  • 22e5ae39d0 add type switch to weight tensor (#16543) Xi Wang 2019-10-28 03:37:55 +08:00
  • d12e674e58 fixing broken links in multiple files - round 3 (#16634) Talia 2019-10-27 12:33:31 -07:00
  • 84d61a1df3 [Perl] - ndarray to native array conversion fix (#16635) Robert Stone 2019-10-27 12:31:50 -07:00
  • 73c6b4a78f fix bad encode (#16641) Hu Shiwen 2019-10-28 03:31:07 +08:00
  • 9f21cddb3f RNNOp only call cuda/cudnn if GPU ctx is requested (#16632) Leonard Lausen 2019-10-27 10:21:14 -07:00
  • c130cc9ef9 add npx reshape (#16640) Xingjian Shi 2019-10-26 18:47:09 -07:00
  • 29e467be25 Move some subgraph verbose to MXNET_SUBGRAPH_VERBOSE=2 (#16622) Zhennan Qin 2019-10-26 15:56:35 +08:00
  • e262455161 Fix index overflow bug in einsum (#16589) Haozheng Fan 2019-10-26 15:43:08 +08:00
  • 8c44af4eba [Gluon] Don't serialize shared parameters twice (#16582) Leonard Lausen 2019-10-26 00:13:42 -07:00
  • 0712f00a2a Faster general take (#16615) Brenton Chu 2019-10-25 20:27:10 -07:00
  • 78627387d4 Fix python doc build issue (#16630) Chaitanya Prakash Bapat 2019-10-25 15:55:50 -07:00
  • c574067d75 Imagenet inference to nightly fix (#16599) Chaitanya Prakash Bapat 2019-10-25 12:36:52 -07:00
  • c0e616f8f4 C Api for simplebind, fix comment for trigoops, add atol to assert (#16585) Chaitanya Prakash Bapat 2019-10-25 09:54:41 -07:00
  • 4e03e6ac20 Disables test_bulking_operator_gpu due to flakiness (#16611) Chaitanya Prakash Bapat 2019-10-24 20:30:52 -07:00
  • ef56334485 julia/docs: more DRY on page rendering (#16396) Iblis Lin 2019-10-25 00:12:15 +08:00
  • 9c99bf2d3e Fix for wrong reqs set after switching from training to inference (#16553) Przemyslaw Tredak 2019-10-24 09:10:48 -07:00