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

Commit Graph

  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • | *
  • * |
  • |/
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • *
  • 55efc8e1 feat(mgb/gopt): add reformat emitter by Megvii Engine Team 2021-07-01 20:31:58 +0800
  • c9d06030 feat(dnn/common): add named tensor shape by Megvii Engine Team 2021-06-30 20:03:02 +0800
  • ff0e6be7 fix(dnn/cuda): fix cutlass tensorop kernels by Megvii Engine Team 2021-08-16 21:43:11 +0800
  • 33676125 feat(dnn/cuda): add tensorcore matmul for fp16 data type by Megvii Engine Team 2021-07-13 17:15:35 +0800
  • 12cdbddd fix(ci): clean fastrun cache in windows and macos ci by Megvii Engine Team 2021-08-20 18:14:23 +0800
  • 31705913 fix(ci): set MGE_FASTRUN_CACHE_TYPE=FILE in ci env by Megvii Engine Team 2021-08-19 18:34:19 +0800
  • f814a4ae ci(mge): update test script by huangxinda 2021-08-20 15:04:55 +0800
  • 2c4ee992 fix(dnn): short cutlass filename in windows by Megvii Engine Team 2021-08-04 11:23:52 +0800
  • b17b56f3 fix(build): fix copy bara error by Megvii Engine Team 2021-08-19 13:33:00 +0800
  • 3c6665f7 feat(lite/whl): merge lite whl to main package by Megvii Engine Team 2021-08-17 22:16:17 +0800
  • 989fdde2 refactor(subgraph): use graph queue to cache compiled op graphs by Megvii Engine Team 2021-08-18 13:51:07 +0800
  • a7a3bf2d test(subgraph): simple test for subgraph by Megvii Engine Team 2021-08-17 20:33:11 +0800
  • d063d577 perf(functional): use fma to reduce elemwise but disable subgraph compilation by Megvii Engine Team 2021-08-11 20:15:02 +0800
  • 2a063f8e fix(subgraph): fix scope mismatch of subgraph content by Megvii Engine Team 2021-08-09 17:28:45 +0800
  • 3206af9d perf(functional/matmul): reimplement matmul with subgraph by Megvii Engine Team 2021-08-06 16:20:12 +0800
  • 8c47c1f1 perf(syncbn): reimplement with subgraph by Megvii Engine Team 2021-08-02 17:29:50 +0800
  • 53da5c79 feat(cg): add comp_seq_sync_device option by Megvii Engine Team 2021-08-04 19:10:40 +0800
  • e1c7b22f perf(ops): enable memory forward for reduce in special cases by Megvii Engine Team 2021-08-11 20:16:47 +0800
  • cd60d268 perf(ops): specialize Broadcast by Megvii Engine Team 2021-08-02 19:11:20 +0800
  • 3fd3e000 feat(ops): add serval utility ops by Megvii Engine Team 2021-08-02 17:20:23 +0800
  • 5b4f7c5d perf(interpreter): unwind ops with make_forward_graph by Megvii Engine Team 2021-08-02 16:54:15 +0800
  • 5798f6ce feat(subgraph): add OpMeth make_forward_graph by Megvii Engine Team 2021-08-02 16:40:05 +0800
  • 48db45d1 perf(interpreter): try put device value with host to reduce d2h by Megvii Engine Team 2021-08-02 16:04:36 +0800
  • a605f38b refactor(opmeth): add OpMethCache struct by Megvii Engine Team 2021-08-02 15:56:17 +0800
  • 0213dbe5 feat(subgraph): add graph builder by Megvii Engine Team 2021-08-02 14:49:57 +0800
  • 0b8dc2c9 refactor(subgraph): add generic encoded_graph by Megvii Engine Team 2021-08-02 14:45:45 +0800
  • 88b3c842 refactor(subgraph): move to subgraph.h by Megvii Engine Team 2021-07-30 19:37:00 +0800
  • 43a9e6e3 fix(third-party): extra logs by Megvii Engine Team 2021-08-18 14:32:26 +0800
  • 43259237 build(dnn/cuda): fix cmake compile dependency for cutlass kernels by Megvii Engine Team 2021-08-17 17:50:30 +0800
  • 1e3af4dd fix(mgb/comp_node): add more info in `comp_node.to_string()` by Megvii Engine Team 2021-08-16 14:12:34 +0800
  • 31031804 fix(mge/optimizer): only disable convert inputs in build-in optimizers by Megvii Engine Team 2021-08-12 19:11:15 +0800
  • fd24dc8e feat(mge/hub/fetcher): add checking of mge version by Megvii Engine Team 2021-08-16 18:23:53 +0800
  • 0ef5183c perf(opdef/reshape): specialize Reshape by Megvii Engine Team 2021-08-11 14:12:54 +0800
  • 77309609 perf(functional/dropout): add fastpath for dropout by Megvii Engine Team 2021-08-11 14:11:44 +0800
  • cc07b96f perf(dnn/relayout): disable copy_last_contiguous when contiguous_size is small by Megvii Engine Team 2021-08-11 14:11:00 +0800
  • b2918f40 feat(third_party): sync patch of cpuinfo and prepare.sh by Megvii Engine Team 2021-08-17 13:20:20 +0800
  • bc2b8d6f fix(windows/fix-prepare): fix prepare on windows by Megvii Engine Team 2020-09-17 14:30:55 +0800
  • ace020c5 fix(git): make git version check work by Megvii Engine Team 2020-08-25 10:44:32 +0800
  • 780fe022 fix(ci): git set user and email by Megvii Engine Team 2021-06-24 14:20:40 +0800
  • d195fdec refactor(mgb): refactor has-usable-algo function for global optimizer by Megvii Engine Team 2021-08-13 19:11:15 +0800
  • d6e50b2c fix(mgb/load-and-run): correct usage string for nchw88 and nchw44-dot by Megvii Engine Team 2021-08-12 13:11:59 +0800
  • 92079e10 docs(mge/distributed): update distributed collective communication Google style docs by Megvii Engine Team 2021-08-05 16:00:48 +0800
  • f9719828 feat(lite): add header licence by Megvii Engine Team 2021-08-04 20:00:37 +0800
  • 71230e9a feat(lite): open source for lite by Megvii Engine Team 2021-08-04 19:24:03 +0800
  • 5fe789ab fix(mge/distributed): skip check when machine_ranks is None by Megvii Engine Team 2021-08-03 14:52:31 +0800
  • 298c4443 fix(mge/module): fix mistake warning message of module auto naming by Megvii Engine Team 2021-08-13 11:04:26 +0800
  • c54a731b fix(utils/network): fix replace oprs by Megvii Engine Team 2021-08-10 17:41:03 +0800
  • 604bb2a5 feat(mgb/dnn): add int atomic add for megdnn by Megvii Engine Team 2021-07-22 16:45:06 +0800
  • 4b08e79d fix(mgb): fix fastrun no_profiling_on_shape_change by Megvii Engine Team 2021-08-04 19:31:43 +0800
  • eab6afab feat(mgb): add padding opr for megbrain by Megvii Engine Team 2021-06-09 18:55:40 +0800
  • 2224a252 fix(mge/opr): add opr_footprint support for PoolingBackward by Megvii Engine Team 2021-08-12 10:26:14 +0800
  • 91675a71 fix(mgb): change span-too-large warning to debug level by Megvii Engine Team 2021-08-06 17:56:45 +0800
  • 7b34936a ci(yaml): fix workflow job trigger event error by luzzyzhang 2021-08-16 13:20:39 +0800
  • cbf3bb18 ci(yaml): fix only pull request run ci lint by luzzyzhang 2021-08-12 21:27:17 +0800
  • 494f23db ci(lint): add commit lint and format check by luzzyzhang 2021-07-21 12:09:29 +0800
  • d9315b7c feat(mge/third_party): update cutlass version by huangxinda 2021-08-12 10:49:22 +0800
  • 0096b45e feat(third_party): add json by Xinda Huang 2021-08-11 12:51:29 +0800
  • 78ebde12 ci(mge): send dingding when tests fail by huangxinda 2021-08-09 22:48:47 +0800
  • ebb7f5ca fix(imperative/python/test): add device judgment for the test case of conv-transpose2d-quantized by Megvii Engine Team 2021-08-10 17:39:11 +0800
  • 66c18f60 fix(ci): fix bazel compile error in new macos by Megvii Engine Team 2021-08-10 14:06:57 +0800
  • a88ecf47 feat(debug): change megbrain do not catch SEGV signal by default change to enable SEGV catch when set MGB_REGISTER_SEGV_HANDLER env only take effect on Linux and MacOS by Megvii Engine Team 2021-08-09 19:03:01 +0800
  • 2950dd8d test(imperative): test autodiff.Function with non tensor arguments by Megvii Engine Team 2021-08-10 14:25:11 +0800
  • 20e8541b refactor(imperative): bind fallback impl on first op method call by Megvii Engine Team 2021-08-09 17:01:03 +0800
  • 18274e02 fix(imperative): fix error message when applying custom function with non-tensor arguments by Megvii Engine Team 2021-08-06 13:39:23 +0800
  • 3e58cbb8 fix(mge/quantization): fix lsq init format error by Megvii Engine Team 2021-08-06 17:10:22 +0800
  • c88a4e5b fix(mgb): fix get env macro by Megvii Engine Team 2021-08-09 14:44:32 +0800
  • b3d2a632 fix(sublinear): fix lb_memory in sublinear by Megvii Engine Team 2021-07-30 18:47:56 +0800
  • 9b4b910d feat(dnn/cuda): integrate cutlass operation table and replace all cutlass wrappers by Megvii Engine Team 2021-07-12 18:17:50 +0800
  • b18feaab feat(dnn/cuda): use cutlass remove shared load imma conv kernel by Megvii Engine Team 2021-07-06 16:33:46 +0800
  • 6b843ccd fix(mge): macos whl build by Tricster 2021-05-06 21:36:45 +0800
  • b74afde8 feat(mgb/opr): let reduce support empty IO by Megvii Engine Team 2021-07-16 18:50:35 +0800
  • 1af350c6 feat(dnn): add fill kernel by Megvii Engine Team 2021-07-16 18:40:03 +0800
  • bbaf524f feat(ci/windows): fix windows ci stability by Megvii Engine Team 2021-08-06 13:35:12 +0800
  • ce2f620e feat(cmake/windows/xp/sp2/inference): implement inference on windows xp (os vesion >= sp2) build with cmake by Megvii Engine Team 2021-08-04 17:59:22 +0800
  • 3eb0505f feat(imperative): add support for quantized conv transpose2d by Megvii Engine Team 2021-07-23 13:45:25 +0800
  • a8309889 test(mge/utils): cover all test data by Megvii Engine Team 2021-08-02 12:48:42 +0800
  • dd1fecdf feat(mge/opr): add cumsum by Megvii Engine Team 2021-07-30 17:33:30 +0800
  • a0c7e047 fix(extern_c_opr): let megbrain throw a exception when load extern c opr fails by Megvii Engine Team 2021-08-03 15:23:46 +0800
  • c68e6695 feat(bazel/windows/xp/sp2/inference): implement inference on windows xp (os vesion >= sp2) build with bazel by Megvii Engine Team 2021-08-02 14:30:30 +0800
  • 125406e3 feat(cmake/windows/cuda): upgrade windows llvm to lastest 12.0.1, for fit cutlass upgrade by Megvii Engine Team 2021-08-02 20:04:55 +0800
  • b8ea6392 (tag: v1.5.0) ci(mge): merge try-import automatically by huangxinda 2021-08-02 14:47:37 +0800
  • 59853922 (release-1.5) chore(release): bump version by Xu Xinran 2021-07-31 23:21:05 +0800
  • 3b452d8c feat(mgb): cuda conv support nhwc format and fp16 dtype by Megvii Engine Team 2021-07-28 19:14:26 +0800
  • 10bcf757 feat(dnn/x86): add algo for x86 max pooling for Window size bigger than 10 and S1 under NCHW88 by Megvii Engine Team 2021-07-27 17:25:30 +0800
  • ddba5c96 fix(core): fix nr_threads is zero by Megvii Engine Team 2021-07-29 14:48:55 +0800
  • 67f11788 perf(arm_common): add elemwise unary multithread support by Megvii Engine Team 2021-07-27 15:52:59 +0800
  • 3afa3893 perf(arm_common): optimize arm common pooling 9x9 and 13x13 by Megvii Engine Team 2021-07-27 13:15:25 +0800
  • d16c5caf fix(mge/dump): fix dump device error with const by Megvii Engine Team 2021-07-23 19:57:05 +0800
  • 2c4ff543 fix(mgb): fix cudnn ConvolutionBackwardData by Megvii Engine Team 2021-07-27 20:01:06 +0800
  • 7138e4fd feat(docs): add docs for megengine.functional.full by Megvii Engine Team 2021-07-23 18:51:53 +0800
  • 0b4a7679 feat(mge/distributed): enable uint8 for collective communication by Megvii Engine Team 2021-07-22 18:56:33 +0800
  • a22b2cf4 ci(copybara): add config files and fix format script by Megvii Engine Team 2021-07-21 14:21:04 +0800
  • 287cab49 fix(mgb/sereg): fix rng operator compatibility by Megvii Engine Team 2021-07-26 17:47:18 +0800
  • e3fc7836 fix(mgb/opr): fix nvof shape error by Megvii Engine Team 2021-07-14 19:34:57 +0800
  • 3f3a256e fix(mge/functional): fix conv* dtype promotion by Megvii Engine Team 2021-07-26 20:44:38 +0800
  • 536506c3 feat(functional): let interpolate support more modes by Megvii Engine Team 2021-07-23 19:33:52 +0800
  • d811dc54 docs(mge/distributed): add document for distributed.backend by Megvii Engine Team 2021-07-21 16:52:42 +0800
  • 9526ee52 docs(distributed.functional): add return type for all_reduce_min by Megvii Engine Team 2021-07-22 17:13:20 +0800
  • 2aba0378 refactor(mgb/dnn): fix group conv is_available by Megvii Engine Team 2021-07-08 15:28:47 +0800
  • 4a92346b refactor(mgb): refactor group conv3d by Megvii Engine Team 2021-07-07 11:29:20 +0800