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

Commit Graph

  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • | *
  • | *
  • | *
  • | *
  • | *
  • | *
  • * |
  • * |
  • * |
  • * |
  • * |
  • * |
  • |/
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • 5507a29b feat(mge/functional): add normalize opr by Megvii Engine Team 2020-04-22 10:14:49 +0800
  • 000663c3 feat(mge/functional): add isnan and isinf oprs by Megvii Engine Team 2020-04-21 12:11:57 +0800
  • ccc95ad2 fix(mgb/gopt): fix remove redundant typecvt pass by Megvii Engine Team 2020-04-23 14:37:36 +0800
  • 9b413219 fix(mge/functional): support negative axis in math.py by Megvii Engine Team 2020-04-22 16:50:17 +0800
  • 207527d1 fix(mge/clamp): fix `F.clamp` by Megvii Engine Team 2020-04-17 15:26:49 +0800
  • 8ac73333 fix(dnn): fix data type check for quantized convolution by Megvii Engine Team 2020-03-06 18:43:19 +0800
  • c59be192 feat(dnn/arm_common/elemwise): add arm_common support chw44 elemwise by Megvii Engine Team 2020-04-17 15:13:17 +0800
  • 6ffcfb4c refactor(mge/dataloader): make plasma_store starting more robust by Megvii Engine Team 2020-04-21 13:21:21 +0800
  • 920eeeb5 fix(mge/build): remove MEGDNN flags from command line by Megvii Engine Team 2020-04-16 13:28:31 +0800
  • a6253fb7 build: remove useless files by Xinran Xu 2020-04-30 19:21:13 +0800
  • 8fcc0727 ci(workflow): add gpu ci workflow by luzzyzhang 2020-04-22 15:28:36 +0800
  • 16f052e9 fix(megdnn): change ver 60 to use cuda capability 50 by luzzyzhang 2020-04-21 17:22:39 +0800
  • c4dfdbd2 refactor(dnn/fallback): refactor im2col by Megvii Engine Team 2020-03-27 15:34:24 +0800
  • 86a3445e build(python_module): use consistent flag declaration in SWIG and c++ by Megvii Engine Team 2020-04-21 01:27:21 +0800
  • e729006f docs(mge/python): add doc for zero_grad by Megvii Engine Team 2020-04-20 15:28:42 +0800
  • bb92ee26 fix(mge/functional): refine doc and add test for linspace by Megvii Engine Team 2020-04-20 13:20:12 +0800
  • ce30d045 feat(mge/functional): add arange in tensor.py by Megvii Engine Team 2020-04-16 14:26:58 +0800
  • 23478a0d test(dnn/cuda): fix cuda int8 test on sm60 by Megvii Engine Team 2020-04-16 15:42:45 +0800
  • cec7fd3c fix(mgb/build): fix multi-machine macro and add test_distributed by Megvii Engine Team 2020-04-20 14:06:37 +0800
  • 3fce8544 refactor(mgb/parampacksplit): remove input offsets on device by Megvii Engine Team 2020-04-17 15:58:29 +0800
  • 11388e58 refactor(mgb/parampacksplit): remove m_mem_plan_fwd by Megvii Engine Team 2020-04-17 15:10:23 +0800
  • d929dd0a docs(mgb/parampack): add description of parampack opr param by Megvii Engine Team 2020-04-14 17:55:51 +0800
  • 5121626e refactor(dnn/parampack): remove parampacksplit by Megvii Engine Team 2020-04-13 11:32:52 +0800
  • 37b67c9b refactor(dnn/parampack): reduce param pack memory use by Megvii Engine Team 2020-04-10 18:21:45 +0800
  • b708f15d refactor(mgb/param_pack): use shared mem for param pack by Megvii Engine Team 2020-04-08 10:58:26 +0800
  • f18259d7 refactor(mge/zmqrpc): zmqrpc return real port or -1 by Megvii Engine Team 2020-04-20 15:06:40 +0800
  • 99fae95e feat(mge/parampack): add user-defined key to pack params by Megvii Engine Team 2020-04-20 16:58:21 +0800
  • 0668b343 feat(dnn/arm_common): add arm_common int8 channel wise stride 1 by Megvii Engine Team 2020-04-07 21:54:27 +0800
  • 6e9a104d fix(mge/cmake): fix cmake for opencl and add some minor fixes by Megvii Engine Team 2020-04-12 10:23:11 +0800
  • 9a42c636 fix(mge/module): add the more warnings in load_state_dict by Megvii Engine Team 2020-04-16 11:23:04 +0800
  • 9b74e6fc fix(mge/random): add one assert in distribution.py by Megvii Engine Team 2020-04-16 18:13:19 +0800
  • fc4dedc0 fix(mge/random): add lower bound and higher bound for uniform sampling by Megvii Engine Team 2020-04-16 15:47:41 +0800
  • 712b87c8 feat(mgb/core): add comp node for cambricon by Megvii Engine Team 2020-02-17 13:22:06 +0800
  • 0b1d516e (tag: v0.3.4, release-0.3) chore(release): bump version by Xinran Xu 2020-05-01 17:23:51 +0800
  • f0959dc9 refactor(mge/examples): add example to use modified sublinear API by Megvii Engine Team 2020-04-26 17:38:06 +0800
  • a2f0e878 feat(mge/api): expose sublinear related parameters at mge api level by Megvii Engine Team 2020-04-22 14:30:27 +0800
  • f44d2632 chore(release): bump version by Xinran Xu 2020-04-22 19:16:17 +0800
  • d84dc7f7 build(python_module): use consistent flag declaration in SWIG and c++ by Megvii Engine Team 2020-04-21 01:27:21 +0800
  • 03728d45 fix(mgb/build): fix multi-machine macro and add test_distributed by Megvii Engine Team 2020-04-20 14:06:37 +0800
  • 856ef627 feat(mgb/core): support copy DeviceTensorND from cpu to cuda by Megvii Engine Team 2020-04-08 23:22:34 +0800
  • ca811c2c fix(mgb/opr/reduce): fix grad function of reduce mean and add testcase by Megvii Engine Team 2020-04-09 19:13:14 +0800
  • ec9de227 fix(mgb/dnn): fix opencl local opr by Megvii Engine Team 2020-04-14 18:01:24 +0800
  • f5833a52 fix(dnn/cuda): fix cublas matmul on sm60 by Megvii Engine Team 2020-04-15 15:19:34 +0800
  • 6bd09b38 ci: refactor run cpp test script by luzzyzhang 2020-04-16 10:40:36 +0800
  • 96742c19 ci: add github action by luzzyzhang 2020-03-30 18:36:19 +0800
  • d74a4ee9 (tag: v0.3.2) fix(mge/build): use consistent genfiles/megbrain_build_config.h by Megvii Engine Team 2020-04-07 16:49:54 +0800
  • 720b6b49 build(mge): copy file required by setup.py to build directory by Megvii Engine Team 2020-04-15 09:36:53 +0800
  • e8494f72 fix(unittest): remove the torch dependency in test_conv by Megvii Engine Team 2020-04-14 20:46:07 +0800
  • 779ed45d docs(mge/parampack): refine parampack doc by Megvii Engine Team 2020-04-14 10:50:03 +0800
  • 8e879404 chore(release): bump version. by Megvii Engine Team 2020-04-14 16:12:57 +0800
  • 0a38906b doc(mge): add openi msg by wenjuan 2020-04-10 09:32:31 +0800
  • 7f60f27a doc(mge): modify readme by wenjuan 2020-04-09 09:57:51 +0800
  • 7ed98dd5 feat(mge/functional): add conv_transpose2d support by Megvii Engine Team 2020-03-31 14:17:45 +0800
  • 20332705 docs(mge/functional): refine tensor transpose doc by Megvii Engine Team 2020-04-08 21:57:17 +0800
  • 97912262 feat(dnn): add nchw44 q8 direct conv stride 1 by Megvii Engine Team 2020-04-09 14:25:50 +0800
  • 944a125c fix(mgb): improve exception safety with smart pointers by Megvii Engine Team 2020-04-09 16:34:45 +0800
  • 32cd9ca0 fix(mgb): remove unnecessary null pointer checks by Megvii Engine Team 2020-04-09 16:34:26 +0800
  • 4abe026b Merge branch 'master' into dev by Megvii Engine Team 2020-04-10 17:30:31 +0800
  • 044695aa feat(mgb/plugin): add profiler for dynamic graph by Megvii Engine Team 2020-04-02 16:36:06 +0800
  • 8b0315b3 fix(mgb): fix nhwcd4 optpass by Megvii Engine Team 2020-04-09 23:44:35 +0800
  • b588d93e fix(mge): fix fast-run cache mismatch by Megvii Engine Team 2020-04-08 18:24:41 +0800
  • 90ca8554 feat(dnn/x86): add avx2 int8 stride1 chanwise multithread conv by Megvii Engine Team 2020-04-09 17:56:04 +0800
  • 0bdb64c5 ci(test): dump test time result to json file by Megvii Engine Team 2020-04-07 15:02:50 +0800
  • 765c9d35 fix(mgb/trt): change log level of trt log by Megvii Engine Team 2020-04-08 16:29:02 +0800
  • dbe2b893 fix(midout): fix brain opr midout 2/2 (also see a6aa1574) fix extern_c_opr midout by Megvii Engine Team 2020-04-03 16:00:40 +0800
  • 888895e9 refactor(mge/test): make test_parampack more stable by Megvii Engine Team 2020-04-08 10:45:24 +0800
  • da522568 fix(mge/data): process mnist without generate new files by Megvii Engine Team 2020-04-07 17:18:11 +0800
  • 538d3de9 feat(dnn/aarch64): add arm64 nchw44 mk matmul by Megvii Engine Team 2020-03-30 22:13:40 +0800
  • 51786484 fix(mge/doc): fix batch_norm2d doc by Megvii Engine Team 2020-04-07 10:50:10 +0800
  • 5e8aa333 refactor(dnn): refactor winograd output transpose by Megvii Engine Team 2020-03-24 20:45:58 +0800
  • c6eb2e8d refactor(dnn): optimize winograd input transpose by Megvii Engine Team 2020-03-23 16:29:03 +0800
  • f077a529 fix(mge/param_pack): release old parameters by Megvii Engine Team 2020-04-03 18:35:28 +0800
  • 48822557 docs(mge/bn): fix docs and tests of batchnorm by Megvii Engine Team 2020-04-03 15:58:31 +0800
  • 2c4d1afe test(mge): move `NUMBER` config to pytest.init by Megvii Engine Team 2020-04-03 15:32:34 +0800
  • f33a92d6 docs(mge): fix api docs of loss and tranform by Megvii Engine Team 2020-04-03 14:26:04 +0800
  • fd83e025 fix(mge/test): fix reproducibility of test_correctness.py by Megvii Engine Team 2020-04-02 16:40:32 +0800
  • b8d27934 fix(mgb/core): fix opr shallow copy in sublinear by Megvii Engine Team 2020-04-03 11:29:08 +0800
  • 075c4562 fix(mgb/core): fix stream propagation in seq_comp_node_opt by Megvii Engine Team 2020-04-03 11:21:29 +0800
  • dc6f89f2 refactor(dnn): refactor winograd strategy helper by Megvii Engine Team 2020-03-31 16:23:51 +0800
  • e1b2d31d test(mge): skip some doctests having different results on gpu by Megvii Engine Team 2020-04-02 17:38:57 +0800
  • 20b67b29 ci(python): add format and lint check for python by Megvii Engine Team 2020-03-30 16:36:11 +0800
  • 62b0c6cd style(mge): apply format.sh by Megvii Engine Team 2020-03-30 18:14:13 +0800
  • fc6aa12e feat(mge):add qint4/quint4 to python by Megvii Engine Team 2020-03-25 16:08:51 +0800
  • a94fb7b1 fix(mgb/gopt): fix convert format opt pass by ensuring the replacement won't change output var's channel by Megvii Engine Team 2020-03-27 11:03:44 +0800
  • 2c2caf33 refactor(mge/data/dataloader): refactor the implementation of parallel dataloader by Megvii Engine Team 2020-03-27 10:44:48 +0800
  • 364dafcc fix typo by Megvii Engine Team 2020-04-01 16:40:25 +0800
  • 94a76969 refactor ImageNet by Megvii Engine Team 2020-04-01 11:04:48 +0800
  • e677ffc8 fix(mge/dataset) fix imagenet dir error by Megvii Engine Team 2020-03-31 15:16:29 +0800
  • c9830d20 fix(mge/test): fix tolerance by Megvii Engine Team 2020-04-01 16:13:46 +0800
  • 20450817 fix(mge/sdk): fix README.md and add missed example by Megvii Engine Team 2020-03-31 17:39:46 +0800
  • 2174fd12 refactor(mge/test): make test_converge more stable by Megvii Engine Team 2020-04-01 14:10:53 +0800
  • dd41e2e4 fix(mge/test):fix integration/manual tests by Megvii Engine Team 2020-03-27 11:48:01 +0800
  • 8d2bbf73 fix(mge/functional): support scalar inputs in elemwise functions by Megvii Engine Team 2020-03-27 10:16:59 +0800
  • ea00b57d refactor(mge/module): use mge rng to sample from uniform/normal distribution by Megvii Engine Team 2020-03-23 17:45:07 +0800
  • a1597cfc docs(mge/param_pack): add description of param pack and add param pack in example run_dist by Megvii Engine Team 2020-03-30 18:08:06 +0800
  • 646829b8 ci(pytest): use common method install pip package by Megvii Engine Team 2020-03-27 18:25:54 +0800
  • 9edce7e2 ci(python_test): run python test with cuda by Megvii Engine Team 2020-03-18 17:01:49 +0800
  • 5fe7cd14 fix(mge/jit): fix typo InternalGraphGenerator by Megvii Engine Team 2020-03-28 22:23:44 +0800
  • 80d3a06f fix(mge/param_pack): fix param pack with just one param by Megvii Engine Team 2020-03-27 19:37:12 +0800
  • 876e799e add(mge) add internal check for mge by Megvii Engine Team 2020-03-26 16:42:54 +0800