f467377f56
[contrib][nnpack] remove training-optimized ops ( #2224 )
2018-12-05 21:28:18 -05:00
93949456ee
[RELAY][PASS] Check Positiveness in FoldScaleAxis ( #2220 )
2018-12-04 15:58:16 -05:00
58398d3862
Port from_nnvm to NNVM as to_relay ( #2144 )
2018-12-04 15:54:01 -05:00
990521dd44
[RELAY][PASS] Fix expr subst and CombineParallelConv2D ( #2218 )
2018-12-03 20:46:58 -05:00
52e55baa47
[Relay] Parser Tests ( #2209 )
2018-12-02 18:58:40 -08:00
eb6d64f1b8
[RELAY] bugfix. ( #2215 )
2018-12-01 22:39:57 -08:00
cb06a184f4
[RELAY][OP] end to end support for pad op. ( #2213 )
2018-12-01 22:26:38 -08:00
cf3f5bcec1
[RELAY][PASS] Memorize FoldScaleAxis backward transform result ( #2214 )
2018-12-01 22:23:52 -08:00
6d32037c64
[SCHEDULE] Fix code lowering when loop condition depends on outer axis. ( #2208 )
2018-11-30 16:05:04 -08:00
94acff30e8
added int type axis for relay reduce ops ( #2199 )
2018-11-30 09:11:53 -08:00
d15477cd00
[Relay][Pass] Fold constant tuple ( #2201 )
2018-11-30 09:03:14 -08:00
e37dbd4e4d
[TVM] Fix llvm codegen (div by power of 2) ( #2204 )
2018-11-30 09:01:38 -08:00
04f7a18320
[Relay] Add support for tuple node in operator fusion ( #2187 )
2018-11-29 20:49:58 -08:00
2f1d709f9c
[PASS] InstrumentBoundCheckers pass ( #2079 )
...
The pass which instruments checkers before
memory accesses (load/store).
This allows to handle invalid memory accesses.
The patch is related to issue:
https://discuss.tvm.ai/t/array-bounds-checking/944
2018-11-29 18:31:59 -08:00
2a5656bf80
[Relay] Alter Op Layout ( #2150 )
...
* [RELAY] Finish alter op pass
* [RELAY] AlterOpLayout Pass
* fix broadcast operators
* fix broadcast operators
* fix broadcast operators
* Support concatenate
* address comments
* address comments
* add comments
* rebase
2018-11-29 17:11:15 -08:00
1e578a6ffd
[TVM] Fix segfault for CanonicalSimplify(x % -1) ( #2194 )
2018-11-29 10:07:53 -08:00
8fe38eef59
[Relay]collapse_sum and broadcast_to compute & schedule ( #2180 )
2018-11-29 10:06:29 -08:00
a259b737f5
[BACKEND][CODEGEN] C codegen with tests ( #2161 )
...
* Implement C code generation with tests
* Code cleanup
* Implement C code generation with tests
* Code cleanup
* tabs to spaces
* make lint compliant
* update export_library and reserve unique C keywords
* move ReserveKeywordsAsUnique to codegen_c
* some documentation and code cleanup
* use tvm.contrib.util for tempdir in testcases
2018-11-28 13:32:31 -05:00
b871500834
[Relay]where compute and schedule ( #2179 )
2018-11-28 09:57:40 -08:00
55599b937b
[Relay]resize op compute and schedule ( #2172 )
2018-11-27 17:49:02 -08:00
7af48f1aa2
[RELAY][IR] Introduce IdNode to preserve var id across rewriting ( #2178 )
2018-11-26 23:07:37 -08:00
246a38a1db
[RELAY]full, full_like compute and schedule ( #2170 )
2018-11-26 22:10:31 -08:00
7880b50c7a
[Relay][Pass] Fix CombineParallelConv2D ( #2167 )
2018-11-26 16:26:55 -08:00
db74c997ea
[Relay] Densenet benchmark ( #2154 )
...
* Port densenet to Relay
* Invoke densenet test in __main()__
* Even the spacing in the IR text format tests
* Forgot to import densenet in init
* Correct reference to densenet in test
2018-11-26 14:11:19 -08:00
fe0eac9431
[RELAY]take and transpose comp and schd ( #2135 )
2018-11-26 10:47:08 -08:00
bcacb764c9
[Relay] Register compute and schedule for upsampling, with miscellaneous fixes ( #2171 )
2018-11-26 10:45:08 -08:00
b5e0d79060
[RELAY]sch and compute for reduce ops ( #2091 )
2018-11-25 22:11:33 -08:00
b903834328
[RELAY][OP] Move computes to cxx, enable concat as injective ( #2166 )
2018-11-25 18:27:29 -08:00
d4f8d20e20
[RELAY][FRONTEND] Initial MXNet frontend support. ( #2163 )
2018-11-25 12:54:03 -08:00
0cf3ddf7fc
Relay reshape reshape_like compute and schedule ( #2159 )
2018-11-25 09:01:44 -08:00
ccd33332e4
Compare relay and numpy outputs in graph runtime test ( #2164 )
2018-11-25 08:48:22 -08:00
9473dca266
[Relay][Op] Add compute, schedule, and tests for expand_dims and squeeze ( #2133 )
2018-11-24 19:33:14 -08:00
b1cf70a885
[RELAY]Testing Inception, Squeezenet, VGG port ( #2013 )
2018-11-24 14:44:41 -08:00
53ac89ede7
[RELAY][PASS] CombineParallelConv2D ( #2089 )
2018-11-22 10:41:40 -08:00
34648272bb
[Relay] Port LSTM to Relay for testing ( #2011 )
2018-11-21 14:11:33 -08:00
2849930465
[RELAY]Slice_like support ( #2014 )
2018-11-20 09:20:09 -08:00
401ffe131e
[Relay][Op] Add test for batch_flatten ( #2134 )
...
* Add tests for batch_flatten and softmax
* Softmax is already tested elsewhere
2018-11-19 21:35:21 -08:00
c113712db7
[RELAY][BACKEND] Enable PlanMemory in the graph runtime. ( #2120 )
2018-11-19 14:59:41 -08:00
6edb3564a1
[RELAY]sch & comp for ops in nn.py ( #2092 )
2018-11-19 12:37:46 -08:00
b71edd76bd
Relay Op sprint (part 2) - Level 1 - log_softmax ( #2128 )
2018-11-19 11:23:37 -08:00
7a01476a3b
[HYBRID FRONTEND] Modify hybrid script to new interface; hybrid op supported; enable compilation_database in CMakeList.txt ( #1757 )
2018-11-18 22:52:17 -08:00
a376eb30b4
[SCHEDULE] Fix boundary check ( #2126 )
...
* Fix boundary check
* Add unittest
2018-11-18 11:16:37 -08:00
cb8a70f41f
[Relay] compute & schedule for relu, softmax ( #2127 )
2018-11-18 08:55:15 -08:00
69ad6aed82
[NNPACK] Add check for NNPACK being available (`nnp_initialize()` succeeding) ( #2119 )
...
This fixes issues with failing tests on PowerPC.
2018-11-15 14:01:36 -08:00
fc55f34fec
[NNPACK] temporary disable nnpack test ( #2115 )
2018-11-14 22:21:03 -08:00
415f7c4b51
[RELAY][PASS] Make FoldConst context and target invariant ( #2114 )
2018-11-14 22:01:15 -08:00
629a293abe
[RELAY][PASS] FuseOps, fix input fusion rule for conv2d ( #2110 )
2018-11-14 16:42:23 -08:00
b252160451
[RELAY][PASS] Bind, FoldConstant ( #2100 )
2018-11-14 08:56:40 -08:00
313e1d99a8
[Relay][OP]NMS ( #1929 )
2018-11-13 14:57:16 -08:00
1f2c815671
[RELAY][OP] strided_slice ( #2094 )
2018-11-13 13:32:38 -08:00
hlu1
Wuwei Lin
Jared Roesch
Josh Pollock
Siva
Tianqi Chen
Siju
Sergei Grechanik
masahi
Denis Khalikov
Lianmin Zheng
Pratyush Patel
Steven S. Lyubomirsky
Animesh Jain
Jian Weng
Junru Shao
Yizhi Liu
Andrew Tulloch
Yao Wang