Commit Graph

  • 32a9baac3b Adding user guidelines for using MXNet built with Large Tensor Support (#16894) Rohit Kumar Srivastava 2019-11-27 14:53:00 -08:00
  • 7713a43353 Update TensorRT tutorial to build-from-source. (#14860) Kellen Sunderland 2019-11-27 09:51:00 -08:00
  • a3b0aa486c [Numpy] Fix imperative basic indexing in numpy (#16902) (#16919) Przemyslaw Tredak 2019-11-27 08:21:47 -08:00
  • a98cefc74c [Numpy] Basic indexing in symbolic interface of DeepNumpy (#16621) Xingjian Shi 2019-11-26 22:58:01 -08:00
  • 9a9d5663d2 wip ib/autograd-custom-func Iblis Lin 2019-11-27 10:34:45 +08:00
  • 8f10d55791 [Numpy] Fix imperative basic indexing in numpy (#16902) Xingjian Shi 2019-11-26 15:34:44 -08:00
  • d2d4876f5d Fix memory leak reported by ASAN in NNVM to ONNX conversion (#15516) kevinzh92 2019-11-26 15:29:33 -08:00
  • 6b00b2ca2a Allow loading from model files with empty weights. (#16061) Jiading Gai 2019-11-26 15:26:02 -08:00
  • 121739a165 updating MXNet version to 1.6.0 in base.h for C APIs (#16906) Rohit Kumar Srivastava 2019-11-26 13:04:00 -08:00
  • a11b7eaca6 Try to fix CI (#16908) Xingjian Shi 2019-11-26 11:04:37 -08:00
  • 00f169f751 port shape op to 1.6.x (#16912) Hao Jin 2019-11-26 09:07:49 -08:00
  • 4c41afd1e6 Backport of #16827, #16791 and #16888 to 1.6 branch (#16901) Przemyslaw Tredak 2019-11-26 09:03:15 -08:00
  • c9585bdfb2 Fix the problem in printing feature in c++ API examples : feature_extract (#15686) quyan 2019-11-26 23:24:18 +08:00
  • 6c7ce24725 Revert "Mkldnn fullyConnect bwd bug fix (#16890)" (#16907) Tao Lv 2019-11-26 19:27:57 +08:00
  • f11592da86 Op Unravel_index PR [Numpy] (#16862) Minghao Liu 2019-11-26 06:31:56 +08:00
  • 6bff547465 Add evaluation_loss to the estimator base class. (#16888) liuzh91 2019-11-26 01:34:44 +08:00
  • ca76bf11c8 Upgrade MKL-DNN dependency to v1.1 (#16823) Tao Lv 2019-11-25 19:22:43 +08:00
  • 436967bee9 Mkldnn fullyConnect bwd bug fix (#16890) rongzha1 2019-11-25 19:18:41 +08:00
  • 9b49cfeaf6 try to move shape_array to npx (#16897) Xingjian Shi 2019-11-24 21:44:39 -08:00
  • 11edb70fc9 [Numpy] Add NumPy support for np.linalg.det and np.linalg.slogdet (#16800) vexilligera 2019-11-25 05:47:02 +08:00
  • 85d3ef3a40 Lamb optimizer update (#16715) Rohit Kumar Srivastava 2019-11-23 22:19:07 -08:00
  • 971ea4f0ad Add arange_like to npx (#16883) Xingjian Shi 2019-11-23 06:20:20 -08:00
  • 6c20fb9e6a [Numpy] Implementation npx.{sample}_n (#16876) Xi Wang 2019-11-23 16:17:09 +08:00
  • 20f8bbc2f9 Refactoring names for mxnet version of nnvm to avoid conflicting with the original tvm/nnvm. (#15303) Disi A 2019-11-23 02:59:13 -05:00
  • e147c9483d [Backport][v1.6.x] Fix the wrong result of sum, mean, argmin, argmax when inputs contain inf or nan (#16884) JackieWu 2019-11-23 15:16:30 +08:00
  • 8b71f61ea0 Initial checkin (#16856) (#16872) Jake Lee 2019-11-22 23:15:51 -08:00
  • 7784ac411c Add support for boolean inputs to FusedOp (#16796) (#16892) Przemyslaw Tredak 2019-11-22 23:15:23 -08:00
  • e3a7141c95 Add support for boolean inputs to FusedOp (#16796) Przemyslaw Tredak 2019-11-22 17:18:54 -08:00
  • a47b540e0a multi-precision lamb update operator Rohit Kumar Srivastava 2019-11-02 00:12:20 +00:00
  • e73c186037 Backport #16798, #16836 and #16838 to 1.6 (#16874) Przemyslaw Tredak 2019-11-21 19:25:06 -08:00
  • 200f0ec8ff [v1.6.x] Backport #16837 into v1.6.x (#16847) Wuxun Zhang 2019-11-22 09:39:07 +08:00
  • 33a3af9aa5 Fix test_gluon.py:test_sync_batchnorm when number of GPUS > 4 (#16835) Pedro Larroy 2019-11-21 15:48:41 -08:00
  • 4da14a2238 add op bitwise_or [numpy] (#16801) Yiyan66 2019-11-22 02:41:29 +08:00
  • ece027c12c add numpy op diagflat [numpy] (#16813) Wentao,Xu 2019-11-22 02:40:19 +08:00
  • 530bd27ed2 fix flakiness of test_np_mixed_precision_binary_funcs (#16873) Hao Jin 2019-11-21 01:00:52 -08:00
  • a8b31a239f Fix InferAttr/InferShapeAttr not calling inference for all nodes in a graph (#16836) Przemyslaw Tredak 2019-11-20 20:53:42 -08:00
  • 307037ec43 USE_NVRTC -> ENABLE_CUDA_RTC to fix maven build. Add compile-guard to fusion. (#16838) Dick Carter 2019-11-20 15:43:44 -08:00
  • 95c3d03a13 Enforces NDArray type in get_symbol (#16871) Serge Panev 2019-11-21 00:21:25 +01:00
  • 7c9cb6b7fe Mixed precison binary op backward (use in) for numpy (#16791) Hao Jin 2019-11-20 15:15:58 -08:00
  • ccf0411d45 Fix test_gluon.py:test_sync_batchnorm when number of GPUS > 4 (#16834) Pedro Larroy 2019-11-20 12:56:03 -08:00
  • 61c8bafdcf Add unoptimized symbol to executor for sharing (#16798) Przemyslaw Tredak 2019-11-19 23:24:43 -08:00
  • f1c6880ccc Fix a few np issues (#16849) Hao Jin 2019-11-19 19:36:08 -08:00
  • 55fe7c5b9d Fix IndentationError in setup.py (#16857) Zach Kimberg 2019-11-19 18:51:49 -08:00
  • 98893102ac Initial checkin (#16856) reminisce 2019-11-19 18:10:49 -08:00
  • 60f53ed5b1 [fix] missing input log higher order. (#15331) kshitij12345 2019-11-20 00:49:52 +05:30
  • 169ed69500 Speed fused_op compilation by caching ptx and jit-compiled functions (#16783) Dick Carter 2019-11-12 14:59:06 -08:00
  • 6834d15e28 [Gluon] Improve estimator usability and fix logging logic (#16810) (#16846) Leonard Lausen 2019-11-20 01:35:30 +08:00
  • 4f14bf4fe5 Add large tensor nightly tests for MKL-DNN operators (#16184) Wuxun Zhang 2019-11-19 17:10:00 +08:00
  • e007dcdaa3 adding API doc for Lamb Phase 1 and 2 Rohit Kumar Srivastava 2019-11-15 01:54:16 +00:00
  • 52f8c9b307 fixing base lamb optimizer Rohit Kumar Srivastava 2019-11-13 19:56:38 +00:00
  • 13dcedda05 initial commit lamb optimizer Anirudh Acharya 2019-10-29 22:56:03 +00:00
  • d4bd7cf11a fix atol for test_preloaded_multi_sgd (#16356) Caenorst 2019-10-02 17:47:59 -04:00
  • e7e6532c92 Add fast implementation of LARS (#16122) Caenorst 2019-09-30 17:14:58 -04:00
  • 135c42ca71 Add OP diag [numpy] (#16786) Minghao Liu 2019-11-19 03:36:43 +08:00
  • 94e7ba7698 Python 2 compatibility fix in base.py perdasilva 2019-11-18 14:54:21 +01:00
  • 8cc4c168e7 [Numpy] Random.normal() with backward (#16330) Xi Wang 2019-11-18 14:27:58 +08:00
  • 9e1ab18e27 [Numpy][Operator] 'where' Implementation in MXNet (#16829) Huang, Guangtai 2019-11-18 12:28:02 +08:00
  • 5a97cf5c92 bump up atol for gradient check (#16843) Chaitanya Prakash Bapat 2019-11-16 22:55:04 -08:00
  • 93f338dde1 [Gluon] Improve estimator usability and fix logging logic (#16810) Leonard Lausen 2019-11-16 21:20:26 +08:00
  • 69fd01838b fix expand_dims fall back when input's ndim is 0 (#16837) Wuxun Zhang 2019-11-16 20:20:59 +08:00
  • ec2f3a6110 [Model] Implement Neural Collaborative Filtering with MXNet (#16689) Xinyu Chen 2019-11-16 16:59:47 +08:00
  • 3256131387 refactor and reduce float types for some functions, also add bitwise_xor (#16827) Hao Jin 2019-11-15 22:29:47 -08:00
  • 9abb151398 Backport to 1.6 (#16773, #16781, #16783, #16716, #16699, #16728, #16769, #16792) (#16832) Przemyslaw Tredak 2019-11-15 17:50:38 -08:00
  • 867c98dcbd Fix SliceChannel Type inference (#16748) (#16797) Anirudh Subramanian 2019-11-15 13:39:29 -08:00
  • 4a27b5c32b [Fix] Add ctx to the original ndarray and revise the usage of context to ctx (#16819) Xingjian Shi 2019-11-14 21:12:13 -08:00
  • b972406979 clean TVM (#16814) Xingjian Shi 2019-11-14 16:10:33 -08:00
  • 33602e510b Update TVM submodule (#16777) reminisce 2019-11-14 15:59:39 -08:00
  • 35839bc2d9 Make mrcnn_mask_target arg mask_size a 2d tuple (#16567) Serge Panev 2019-10-23 00:49:50 +02:00
  • 70b65cb37c Load NDArray only to GPU if GPU is present (#16432) Leonard Lausen 2019-10-17 15:17:59 -07:00
  • d1797cf00f Move MRCNNMaskTarget op to contrib (#16486) Serge Panev 2019-10-15 21:01:28 +02:00
  • 7e21bda4ae Fix nightly build (#16773) Haozheng Fan 2019-11-15 05:38:25 +08:00
  • 5139e2cf51 add gammaln, erf, erfinv (#16811) Xi Wang 2019-11-15 04:20:18 +08:00
  • 017f6fa799 Static link MKL-DNN library (#16731) Tao Lv 2019-11-14 09:33:34 +08:00
  • 42d3182e5a migrate cudaMemcpy to cudaMemcpyAsync+cudaStreamSynchronize (#16790) Hao Jin 2019-11-13 15:01:51 -08:00
  • e88e97fa48 [Numpy] Fix collect_params().zero_grad() in gluon numpy interface (#16716) Xingjian Shi 2019-11-13 11:11:40 -08:00
  • 9b25db05b4 Fix numpy-compatible mean output type for integer inputs (#16792) Hao Jin 2019-11-13 11:06:49 -08:00
  • 7d4f2f3b10 [MXNET-1421] Added (CuDNN)BatchNorm operator to the list of mirrored operators (#16022) Bojian Zheng 2019-11-13 12:28:19 -05:00
  • 47fd3a02f4 Link fixes4 (#16764) Talia 2019-11-12 17:58:35 -08:00
  • 2c02bfffb2 Speed fused_op compilation by caching ptx and jit-compiled functions (#16783) Dick Carter 2019-11-12 14:59:06 -08:00
  • e3f5dade60 Fix (#16781) Przemyslaw Tredak 2019-11-12 10:06:59 -08:00
  • 52716deeae [MXNET-1426] Fix the wrong result of sum, mean, argmin, argmax when inputs contain inf or nan (#16234) JackieWu 2019-11-12 23:17:15 +08:00
  • 3885bbed8c [Done] BilinearResize2D optimized (#16292) chinakook 2019-11-12 19:03:10 +08:00
  • 02f4f0503d [Numpy] Add sampling method for bernoulli (#16638) Xi Wang 2019-11-12 17:07:02 +08:00
  • b9b56e6628 fix inv test flakiness using random matrices generated by SVD (#16782) vexilligera 2019-11-12 15:42:07 +08:00
  • 27672b9900 [Perl] - ndarray operator overloading enhancements (#16779) Robert Stone 2019-11-11 13:23:31 -08:00
  • 177fd3e432 Support boolean elemwise/broadcast binary add, multiply and true_divide (#16728) Hao Jin 2019-11-10 20:29:54 -08:00
  • 618ce8b0e3 add numpy op nan_to_num (#16717) alicia 2019-11-11 06:40:35 +08:00
  • 927def2b38 Fix rtrue_divide grad (#16769) reminisce 2019-11-10 13:31:42 -08:00
  • 74529dcd6b [NumPy] NumPy support for linalg.inv (#16730) vexilligera 2019-11-10 18:29:10 +00:00
  • 57b9b2c769 change _generate_op_module_signature get_module_file open with encoding="utf-8",it fix some encode error in Chinese windows system. (#16738) Hu Shiwen 2019-11-10 09:03:03 +08:00
  • b1aba6a8ad Backport of #16711, #16737, #16408 to 1.6 branch (#16763) Przemyslaw Tredak 2019-11-08 15:06:39 -08:00
  • a37dcd46cb Fix SliceChannel Type inference (#16748) Anirudh Subramanian 2019-11-07 18:41:18 -08:00
  • 5dfa1219ef [MKLDNN] use dim_t instead of int in slice/transpose operators (#16737) Wuxun Zhang 2019-11-08 09:30:02 +08:00
  • c38b527843 [MKLDNN] Fix int8 convolution/fc bias overflow (#16734) Zhennan Qin 2019-11-07 12:35:04 +08:00
  • d967be912e [BUG FIX] Always preserve batch dimension in batches returned from dataloader (#16233) 夏鲁豫 2019-11-07 11:10:08 +08:00
  • ecb7a3ada1 Update submodule dmlc-core (#16742) Dick Carter 2019-11-06 18:21:47 -08:00
  • da33da3cb8 Add MXNet Ops for fast multihead attention (#16408) Caenorst 2019-11-06 21:00:43 -05:00
  • c4580ae414 Fix for wrong reqs set after switching from training to inference (#16553) Przemyslaw Tredak 2019-10-24 09:10:48 -07:00
  • 58b824f8c2 fix R docs (#16733) Thomas Delteil 2019-11-06 11:12:37 -08:00
  • 579b9ddd3f fix another typo cfujitsang 2019-11-01 13:22:57 -07:00