wangwei
/
MegEngine
Not watched
1
0
Fork 0
Code
Releases 31 Wiki Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain

Commit Graph

  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • |\
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • | | *
  • | | *
  • | | |\
  • | |_|/
  • |/| |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • * | |
  • 72619bb4 fix(whl/linux): fix mv failed if output have old whl by Megvii Engine Team 2020-11-06 18:10:35 +0800
  • 9682db98 feat(mgb): add jit mlir elemwise broadcast by Megvii Engine Team 2020-10-10 14:28:10 +0800
  • 89303cd8 feat(megdnn/rocm): add bn for rocm backend by Megvii Engine Team 2020-11-04 18:07:41 +0800
  • 7cd846a5 fix(mge/batch_norm): fix batch_norm check when trace(symbolic=True) by Megvii Engine Team 2020-11-06 12:43:45 +0800
  • 120e719e refactor(imperative): refactor OpTrait methods and registration by Megvii Engine Team 2020-11-04 14:47:31 +0800
  • aea829c9 feat(megdnn/rocm): add average inclusive mode for pooling by Megvii Engine Team 2020-11-05 00:45:26 +0800
  • a3550f91 test(rocm): Fix rocm testcase by Megvii Engine Team 2020-11-04 01:01:32 +0800
  • 1e90c457 refactor(mgb/opr): move fastrun out of conv opr by Megvii Engine Team 2020-11-02 22:04:37 +0800
  • 2fc77caf fix(whl/macos/windows/linux): fix macos/windows whl broken caused by b2dbda96b, and add macos/windows whl import test, Integration Test will test linux whl import, so we do not add linux import test at MR test by Megvii Engine Team 2020-11-04 14:44:46 +0800
  • c8920945 ci(mge): remove blank line for windows doctest by Megvii Engine Team 2020-11-05 10:29:22 +0800
  • e082e277 refactor(mge/data): Refactor megeninge.data.dataset by Megvii Engine Team 2020-11-05 20:39:04 +0800
  • f04e0d77 feat(mge/data): dpflow dataset, stream sampler and loader by Megvii Engine Team 2020-10-21 13:46:35 +0800
  • be511a56 fix(mge/imperative): fix tensor astype failed for quantized type by Megvii Engine Team 2020-11-03 11:52:12 +0800
  • b309890c docs(mge): pytest for sphinx docstring by Megvii Engine Team 2020-11-03 18:39:27 +0800
  • 1b568517 fix(mge/trace): fix graph option in trace by Megvii Engine Team 2020-10-30 14:11:01 +0800
  • f2f5f9ac fix(mgb/jit): remove MGE_USE_SYSTEM_LIB from mlir linking by Megvii Engine Team 2020-11-02 17:45:00 +0800
  • 12178011 perf(mge): add opdef for broadcast by Megvii Engine Team 2020-10-29 17:00:02 +0800
  • fccb2510 perf(mge): add opdef for elemwise and batchnorm by Megvii Engine Team 2020-10-27 18:56:21 +0800
  • c008cf37 refactor(mge): delete old dump_with_testcase_mge by Megvii Engine Team 2020-10-27 14:28:57 +0800
  • 2a3f4d09 refactor(dnn/arm): refactor CPU heuristic algo selection by Megvii Engine Team 2020-08-03 22:34:34 +0800
  • 981f487b chore(third_party): update upstream of cutlass by Megvii Engine Team 2020-11-05 17:48:13 +0800
  • 5cfe7365 fix(mgb/core): ensure all VarNodes would be handled in init_ready_event by Megvii Engine Team 2020-10-29 20:04:46 +0800
  • 2d0c9690 fix(opr): add mask check for dct by Megvii Engine Team 2020-10-26 20:00:00 +0800
  • 9ffde212 feat(python): add dct python opr by Megvii Engine Team 2020-10-23 20:01:41 +0800
  • 45032220 feat(opr): add dct opr by Megvii Engine Team 2020-10-23 20:01:41 +0800
  • ba66e1d0 feat(dnn): add nchw_fp32 nchw44_qint8 cuda dct by Megvii Engine Team 2020-10-23 20:01:41 +0800
  • b3278229 feat(mge/grad_manager): add `clear_grad` method for GradManager by Megvii Engine Team 2020-10-29 11:37:57 +0800
  • 2627e1f7 fix(mge/grad_manager): allow multiple calls of `release` by Megvii Engine Team 2020-10-28 10:54:11 +0800
  • 67a543f3 build(mlir): remove unused llvm target to reduce compile time by Megvii Engine Team 2020-10-29 16:21:04 +0800
  • a9f98e9c refactor(meg/internal): move interal codes back to megbrain by Megvii Engine Team 2020-10-22 19:08:18 +0800
  • 65c54913 feat(load_and_run): feat --input parameter by Megvii Engine Team 2020-10-29 14:09:47 +0800
  • 8419a4fb build(third_party): update llvm-project and adapt to mlir interface changes by Megvii Engine Team 2020-10-30 10:00:22 +0800
  • 170897f2 feat(mgb/jit): add more elemwise modes for mlir backend by Megvii Engine Team 2020-10-22 08:27:23 +0800
  • ca717806 fix(mgb): fix wrong use of macro MGB_ENABLE_GRAD by Megvii Engine Team 2020-10-27 19:35:11 +0800
  • e8bf5bc0 fix(mge): ensure contiguous when passing tensor from graph rt to imperative rt by Megvii Engine Team 2020-10-26 21:47:05 +0800
  • 3af10563 feat(mge/grad): attach grad immediately by Megvii Engine Team 2020-09-29 14:55:11 +0800
  • dc3c17ba chore(cmake): fix a cmake typo by Megvii Engine Team 2020-03-26 11:11:07 +0800
  • 44b27f0d build(3516): fix some cpu flags build failed and fix 3516 ycm by Megvii Engine Team 2020-10-26 14:34:44 +0800
  • cab46f2a fix(dnn): fix relayout format when group conv group=1 by Megvii Engine Team 2020-10-21 14:12:54 +0800
  • 8b154fd7 fix(license): correct and format the content of license to make it be detected by GitHub by Megvii Engine Team 2020-10-22 13:26:39 +0800
  • b13c78a5 fix(ios): fix cmake ios cross build by Megvii Engine Team 2020-10-23 10:16:42 +0800
  • cfc430b8 fix(mbg/opr): fix ModifySubtensorImplHelper infer shape depend input value for valid check by Megvii Engine Team 2020-07-17 14:16:07 +0800
  • 8764a6c8 feat(dnn/cuda): add volta dp4a int8 sass kernel by Megvii Engine Team 2020-09-22 14:02:47 +0800
  • e296a684 fix(build/doc): fix build md by Megvii Engine Team 2020-10-21 17:03:31 +0800
  • 3635af62 style(atlas): add comment for async d2d by Megvii Engine Team 2020-10-21 18:07:43 +0800
  • d68d4d1d perf(atlas): use async d2d by Megvii Engine Team 2020-10-09 18:41:24 +0800
  • d8ac6c70 fix(quantize): fix quantize calibration dtype issue by Megvii Engine Team 2020-10-21 16:45:37 +0800
  • 4afa4b72 fix(mgb/oppass): fix fuse conv bias pass when bias is full bias and bias is const provider by Megvii Engine Team 2020-10-15 17:43:41 +0800
  • 50d64bd3 fix(pytest/windows/impertive): open impertive pytest on windows by Megvii Engine Team 2020-10-17 19:41:54 +0800
  • dfc3935d fix(pytest/windows/impertive): fix windows wrong test skip by Megvii Engine Team 2020-10-17 21:30:24 +0800
  • 786399c5 fix(pytest/windows/impertive): fix impertive pytest failed on windows by Megvii Engine Team 2020-10-14 12:11:47 +0800
  • 215f88f3 fix(dnn/argmxx): fix argmxx on inf by Megvii Engine Team 2020-10-15 15:31:44 +0800
  • 03d0cc02 feat(gopt): add remove redundant copy pass by Megvii Engine Team 2020-10-20 00:46:31 +0800
  • 545567d3 fix(atlas): add NO_MEM_RECLAME flag for the output of atlas runtime opr by Megvii Engine Team 2020-10-17 00:14:56 +0800
  • 1fa143ce refactor(mge/functional): matmul supports symbolic shape, batched mv multiply by Megvii Engine Team 2020-10-19 14:16:03 +0800
  • d47cf332 build(third_party): update llvm-project and adapt to mlir interface changes by Megvii Engine Team 2020-09-25 15:40:05 +0800
  • 95f6b531 feat(opr/nvof): add nvof operator by Megvii Engine Team 2020-10-19 13:09:54 +0800
  • 9748aebe refactor(mge): tensor_shape -> symbolic_shape by Megvii Engine Team 2020-10-14 12:53:06 +0800
  • 8acc3acf fix(mge/functional): do not del anything by Megvii Engine Team 2020-10-16 18:36:02 +0800
  • 92b12685 feat(dnn/aarch64): add aarch64 int8X8X16_mk4_k8x8x8 matmul, performance is better by Megvii Engine Team 2020-09-24 18:37:49 +0800
  • 5ee1a1c4 fix(mge/functional): del graph in __init__ by Megvii Engine Team 2020-10-16 16:39:14 +0800
  • 67013463 docs(mge): fix some docstring format problem by Megvii Engine Team 2020-10-13 18:37:23 +0800
  • 5cc043f0 fix(imperative): sync command queue at exit by Megvii Engine Team 2020-10-13 14:48:22 +0800
  • 9dc55386 fix(mgb): update fastrun log level when not set timeout by Megvii Engine Team 2020-10-14 18:01:06 +0800
  • 0075e6ac refactor(mge): refactor cross_entropy by Megvii Engine Team 2020-10-14 15:48:35 +0800
  • f4b16932 test(mgb/imperative): add adaptive pooling pytest by Megvii Engine Team 2020-10-10 11:15:50 +0800
  • 37e56f4b fix(mge/functional): fix clip under trace(symbolic=True) by Megvii Engine Team 2020-10-13 16:10:39 +0800
  • 8230ea2d Update README.md by zhanghaolong 2020-09-17 15:47:33 +0800
  • 74180979 chore(version): bump version by Megvii Engine Team 2020-10-14 20:26:35 +0800
  • e8ea6dd5 fix(windows/fix-prepare): fix prepare on windows by zhanghaolong 2020-09-17 14:30:55 +0800
  • a594f96b (tag: v1.0.0, release-1.0) chore(version): bump version by Megvii Engine Team 2020-10-14 20:26:35 +0800
  • 0a0e4b60 Merge branch 'master' into release-1.0 by Megvii Engine Team 2020-10-14 20:23:39 +0800
  • 8d1f3025 fix(mge): fix batch norm dump by Megvii Engine Team 2020-10-12 21:12:36 +0800
  • 40e778fb fix(mgb): fix-deconv-io16c32-pass-fail by Megvii Engine Team 2020-09-08 19:35:19 +0800
  • 9f4bffbd fix(mge/tensor): fix valid_broadcast by Megvii Engine Team 2020-10-12 17:50:32 +0800
  • af349d61 fix(mge/functional): fix op mismatch when tracing NMSKeep by Megvii Engine Team 2020-09-29 14:48:57 +0800
  • d502e79f feat(mge): make F.norm numpy compatible by Megvii Engine Team 2020-10-12 14:45:16 +0800
  • d31a4fff refactor(mge): use logsumexp in cross_entropy by Megvii Engine Team 2020-10-12 14:17:49 +0800
  • 20e93630 feat(mge): rename batch_norm2d -> batch_norm by Megvii Engine Team 2020-10-10 21:29:17 +0800
  • aa626726 feat(mge): make F.eye numpy compatible by Megvii Engine Team 2020-10-10 21:09:13 +0800
  • 7589bce4 feat(mge): do not export F.nn.sync_batch_norm by Megvii Engine Team 2020-10-10 19:59:26 +0800
  • d4e2d272 feat(mge): do not export F.nn.conv_bias_activation by Megvii Engine Team 2020-10-10 19:56:59 +0800
  • e64760fe docs(mge): warp_perspective, gather & scatter by Megvii Engine Team 2020-10-10 19:48:29 +0800
  • 7d8a7e3e feat(mge): do not export F.loss.* by Megvii Engine Team 2020-10-10 19:40:58 +0800
  • 480375fb feat(mge): rename accuracy -> topk_accuracy by Megvii Engine Team 2020-10-10 19:22:01 +0800
  • 5b2f0129 feat(mge): rename cross_entropy_with_softmax -> cross_entropy by Megvii Engine Team 2020-10-10 19:16:33 +0800
  • cf0507e1 feat(mge): remove F.identity by Megvii Engine Team 2020-10-10 14:37:41 +0800
  • 09241a1f feat(mge): remove param_pack_* from functional by Megvii Engine Team 2020-10-10 14:18:24 +0800
  • 0d4568d6 feat(mge): rename broadcast -> broadcast_to by Megvii Engine Team 2020-10-09 23:30:59 +0800
  • f551d44e feat(mge): rename arange paramter end -> stop by Megvii Engine Team 2020-10-09 23:02:58 +0800
  • 9c5ea689 feat(mge): remove dimshuffle by Megvii Engine Team 2020-10-09 22:59:00 +0800
  • 1e2117f6 feat(mge): remove add_axis and remove_axis by Megvii Engine Team 2020-10-09 22:51:57 +0800
  • 5b7ae268 feat(mge): do not export F.nn.roi_pooling by Megvii Engine Team 2020-10-09 22:35:52 +0800
  • d22d1676 feat(mge): do not export F.nn.roi_align by Megvii Engine Team 2020-10-09 22:34:44 +0800
  • 9c561134 feat(mge): do not export F.nn.nms by Megvii Engine Team 2020-10-09 22:32:58 +0800
  • 56d624f5 feat(mge): rename log_softmax -> logsoftmax by Megvii Engine Team 2020-10-09 22:31:45 +0800
  • 164d9cdf feat(mge): do not export F.nn.linear by Megvii Engine Team 2020-10-09 22:30:26 +0800
  • e42679b5 feat(mge): do not export F.nn.interpolate by Megvii Engine Team 2020-10-09 22:25:28 +0800
  • fd802e97 feat(mge): do not export F.nn.embedding by Megvii Engine Team 2020-10-09 22:24:43 +0800
  • a640eeff feat(mge): remove batched_nms by Megvii Engine Team 2020-10-09 22:13:57 +0800