5d533ec99b
Improve x86 Inception ( #1506 )
...
* Improve x86 pooling and concat
* Fix
* Fix test concatenate correct layout
* Add conditional vectorize
* Fix lint
* Modify schedule for global pooling
* Fix
* Fix warning
* Fix alter layout test
* Remove vectorization for pooling when using 4D layout
* Remove vectorization for 4D concat
* Fix concatenate layout
* Fix concatenate schedule
* Fix concat
* Fix lint
* Fix concat
* Simplify pooling logic
* Update docstring
* Fix test topi pooling
* Small changes
2018-08-20 13:27:31 -07:00
4dc21bdb29
[NNVM][DARKNET]Yolo and Upsample frontend support ( #1501 )
...
* Yolo and Upsample frontend support
* Lint fix
* Mac support added
* Code clean and trigger CI
2018-08-20 10:13:35 -07:00
b00aabc5ca
Add missing check when deciding conv op and injective op are in the same group ( #1622 )
2018-08-19 19:30:01 -07:00
1c66012ac9
fix import ( #1621 )
2018-08-19 19:29:35 -07:00
a5f2759ec5
[FRONTEND][COREML]MultiplyLayerParams L2NormalizeLayerParams and UpsampleLayerParams support … ( #1511 )
2018-08-18 11:20:58 -07:00
9b0e499054
[NNVM] TF: Add Pack operation ( #1570 )
2018-08-17 21:40:52 -07:00
6cd5a8f991
[NNVM] Bug fix Prevent fusing convolution with injective op ( #1608 )
2018-08-16 19:37:56 -07:00
093dc74136
[NNVM] Add ONNX upsample converter ( #1591 )
2018-08-16 16:35:23 -07:00
a80356bbe0
[NNVM] Add symbol for inception v3 ( #1604 )
2018-08-15 15:42:24 -07:00
54a115ef14
[FRONTEND][DARKNET]LSTM and GRU support ( #1576 )
2018-08-14 10:48:02 -07:00
bc7431bda6
[NNVM][POOL] bug fix. Remove the hardcode. ( #1600 )
2018-08-14 09:44:33 -07:00
84296f9632
[FRONTEND][TENSORFLOW] Optimized tensorflow testcases ( #1546 )
...
* [NNVM][TENSORFLOW] Optimized tensorflow testcases
* Replace Constants with Placeholder
* Review comment fix
2018-08-13 14:14:26 -07:00
672147c8e1
add conv2d transpose and fix bugs ( #1566 )
2018-08-09 11:17:37 -07:00
53d243114d
Separate fusion and Compilation ( #1564 )
...
* Separate fusion and compilation
* fix description of graph_fuse.h
* fix lint
* fix @masahi 's comments, move fusion out of target
* fix graph passing and make fused_entries singula in graph attr
* fix typo
* fix some comments
* run test again
* remove rvalue for graphfuse and graphfindfusiablegroups
2018-08-08 13:51:47 -07:00
c9f9a3f9be
l2normalization operator support for tensorflow ( #1528 )
2018-08-08 10:26:10 -07:00
7ea06e6e4c
[ONNX]onnx gather bug fix ( #1543 )
2018-08-08 10:23:51 -07:00
60da47055a
add tvm external registry entrypoint ( #1562 )
2018-08-08 10:11:24 -07:00
1ed28aeb59
[NNVM] Enhance operator fusion for more element wise patterns ( #1548 )
2018-08-08 10:07:44 -07:00
0241fdc5d2
[FRONTEND][ONNX]LRN support for ONNX ( #1518 )
...
* LRN support for ONNX
* [ONNX] Updated lrn testcases
2018-08-06 19:36:37 -07:00
a8574e7bb8
[FRONTEND][TENSORFLOW] Add Pad and PadV2 support ( #1545 )
...
* [FRONTEND][TENSORFLOW] Add Pad and PadV2 support
* Add assettion to _pad, and fix testcase for pad.
2018-08-06 14:07:38 -07:00
7b59b8ef80
[NNVM][TENSORFLOW] Cleanup redundant code. ( #1551 )
2018-08-04 09:08:51 -07:00
7c050e4211
Fix RPC ( #1542 )
2018-08-02 21:47:04 -07:00
0f431fd66d
Fix rpc testcase ( #1538 )
2018-08-02 19:12:36 -07:00
be718ba1d5
Update test_rpc_exec.py
2018-08-02 16:43:23 -07:00
6d1dc4ae5f
[NNVM] Support argmax/argmin in tensorflow frontend ( #1514 )
2018-08-02 09:48:41 -07:00
71cff3e862
[MXNET] LRN support in MXNET frontend ( #1520 )
2018-08-02 09:02:34 -07:00
a2870fefc8
[NNVM][TENSORFLOW]Local Response Normalization added for tensorflow ( #1522 )
2018-08-02 09:01:22 -07:00
32076df815
[AUTOTVM] TOPI integration for ARM CPU ( #1487 )
2018-08-02 08:59:25 -07:00
b625b992a4
keras.layers.ReLU added ( #1530 )
2018-08-02 08:58:24 -07:00
ba4cc7bae4
Onnx Gather operator added ( #1513 )
2018-08-01 14:39:52 -07:00
d6f3bf166b
Add some missing operators ( #1524 )
2018-08-01 12:27:49 -07:00
82e5938514
[NNVM] remove keepdims from expand_like arguments ( #1517 )
2018-08-01 09:59:54 -07:00
3597dfb624
[NNVM] Fix gradients for broadcast_div ( #1512 )
2018-07-31 13:35:38 -07:00
303b00eb42
C-RNN layer support is added ( #1492 )
2018-07-30 07:09:43 -07:00
b65a13ddd2
Activations for coreml added ( #1508 )
2018-07-30 07:09:00 -07:00
325c82740b
[NNVM] Fix check in layout parsing ( #1502 )
...
* [NNVM] Fix check in layout parsing
* add one workload
2018-07-27 20:46:47 -07:00
6931274420
[NNVM] Fix grads for sum and expand_like ( #1455 )
2018-07-27 08:50:49 -07:00
cf9db7ea66
[NNVM] Add argmax and argmin operations from topi ( #1462 )
2018-07-25 12:08:39 -07:00
a808a9871e
[NNVM][TENSORFLOW] LSTM operator and PTB word prediction frontend ( #1389 )
2018-07-25 09:56:05 -07:00
55a08deca0
[NNVM][DARKNET]Logistic activation added ( #1477 )
2018-07-25 09:20:46 -07:00
c3b462fcf4
Fix GetReduceAces ( #1460 )
2018-07-20 08:54:47 -07:00
df8e338201
[DARKNET]RNN Support for darknet ( #1443 )
2018-07-19 10:20:21 -07:00
ca5397d577
Tutorial enhancement to keep it clean on docs.tvm.ai ( #1450 )
2018-07-19 09:58:12 -07:00
72fa4c1d93
[NODE][REFLECTION] Support NDArray as field ( #1452 )
2018-07-18 16:18:58 -07:00
8dbe779466
Fix inceptionv3 ( #1446 )
2018-07-18 09:08:07 -07:00
4e29ec334c
[NNVM][TENSORFLOW] bug fix on bilinear and resize op integration in frontend. ( #1440 )
2018-07-18 09:04:48 -07:00
2a1c2ba9bc
Fix conda package builds ( #1445 )
2018-07-17 13:45:08 -07:00
c19cf6f734
[NNVM] Add symbol squeezenet ( #1436 )
2018-07-15 15:23:18 -07:00
b5ff88b0cd
[Darknet] softmax temperature in frontend ( #1429 )
2018-07-13 09:52:14 -07:00
1fb2d7e22e
Add support for absolute opeartion ( #1406 )
2018-07-13 09:49:49 -07:00
ee3c1b09b9
[TOPI]Add where operator ( #1416 )
2018-07-12 20:04:41 -07:00
81db22c56a
[RPC] graduate tvm.contrib.rpc -> tvm.rpc ( #1410 )
2018-07-09 15:23:41 -07:00
0ec70800dd
fix CorrectLayout for softmax & log_softmax ( #1401 )
2018-07-09 13:19:54 -07:00
badcb6302b
change line seperator of tensorflow/test_forward from CRLF to LF ( #1405 )
2018-07-09 09:54:53 -07:00
b30ae8ac6d
[TOPI][DARKNET]Yolo op added ( #1372 )
2018-07-08 21:50:52 -07:00
4e7b548efc
[DARKNET FRONTEND]Batchnorm added as part of Dense op for running rnn model for next wo… ( #1385 )
2018-07-08 21:50:28 -07:00
05f4362e8a
[NNVM][Keras] allow only tensorflow backend ( #1392 )
2018-07-08 10:48:26 -07:00
3a0b757c10
[NNVM][TOP] broadcast versions corresponding to topi: mod, max, min, pow, left_shift, right_shift greater, less, equal, not_equal, greater_equal and less_equal. ( #1383 )
2018-07-07 18:51:57 -07:00
4dd41ddb5b
Move SGX enclave signing key to gist ( #1393 )
2018-07-06 10:17:52 -07:00
10b7757ab6
[NNVM][TENSORFLOW] Fixed variable ops shape parsing issue ( #1381 )
2018-07-04 20:45:30 -07:00
2fa0eca13a
[NNVM][ONNX] Slice, Floor, Ceil, Clip and MatMul support for frontend #1297 ( #1371 )
2018-07-04 11:31:02 -07:00
d7c600b80a
[TOPI] Update parameter name of conv2d ( #1380 )
2018-07-04 09:04:03 -07:00
30f09bea53
[AVG POOL] Asymmetric padding (SAME) support. ( #1346 )
2018-07-04 09:03:40 -07:00
40ac206438
add take frontend ( #1307 )
2018-07-04 09:02:25 -07:00
4503f77b0a
[NNVM][TENSORFLOW] Sigmoid op support #1367 ( #1369 )
2018-07-03 19:15:02 -07:00
28acb18432
[NNVM] Move FTVMCompute registration of cast, greter, less to C++. ( #1370 )
2018-07-03 09:45:11 -07:00
32a55f8814
Prelu bug fix ( #1358 )
2018-06-30 07:55:15 -07:00
ab0d1862ab
[NNVM] Move FTVMCompute registration of the elementwise operator to c++ ( #1351 )
2018-06-28 21:28:18 -07:00
b154e6b904
[NNVM] Initial mixed precision support of conv2d ( #1356 )
2018-06-28 10:37:04 -07:00
4fb58115ea
Strided_slice added in NNVM ( #1318 )
2018-06-28 08:05:33 -07:00
71f88611cd
Improve schedule load, add slice_like ( #1299 )
2018-06-26 18:42:17 -07:00
50681784f4
[TOPI] Add C++ implementation of elementwise operators ( #1306 )
2018-06-26 18:40:31 -07:00
44d8203fc5
[NNVM][TOPI] Add gradients for broadcast_* ops ( #1234 )
2018-06-26 18:39:45 -07:00
2f77a127be
[NNVM][ONNX] Squeeze and Unsqueese operators. ( #1339 )
2018-06-26 11:59:53 -07:00
113b46ec80
[NNVM][ONNX] Shape operator support (limited/differed) - #1297 ( #1333 )
2018-06-26 10:15:53 -07:00
373a8caa79
[NNVM][TENSORFLOW] Mobilenet support. ( #1335 )
2018-06-26 10:15:24 -07:00
c3df7726ac
support t attr in onnx ( #1300 )
2018-06-26 09:03:52 -07:00
59a8d09936
[NNVM][TOPI] Add FTVMCompute for matmul ( #1239 )
2018-06-25 20:41:01 -07:00
bb7df695cd
[NNVM][CONVOLUTION] Group convolution generalization for NHWC ( #1232 )
2018-06-24 09:09:09 -07:00
5b33e7b898
Windows VS2015 Compile error - 'conversion from 'double' to 'float' requires a narrowing conversion' ( #1329 )
2018-06-24 09:05:10 -07:00
bdd4bab02d
[NNVM] Introduce const shift ops ( #1325 )
2018-06-23 18:57:18 -07:00
f216b25e01
[FRONTEND][MXNET] Add expand_dims supoort ( #1317 )
...
* [FRONTEND][MXNET] Add expand_dims supoort
* fix lint
2018-06-23 08:57:27 -07:00
db4be63c51
[TOPI] Numpy consistency: always broadcast binary op. ( #1321 )
2018-06-22 16:49:07 -07:00
fb88b74e4a
CPP implementation of L2Norm and LRN ops ( #1157 )
2018-06-22 10:06:40 -07:00
464c8c2667
Fix a bug in Symbol::Compose when using subgraphs as input ( #1314 )
2018-06-21 19:25:55 -07:00
baa04599da
[CONTRIB] TVM download utility based on urllib2/urlib.request ( #1313 )
...
moving nnvm/testing/download.py to python/tvm/contrib/download.py to be used as a general TVM download utility
2018-06-21 15:39:43 -07:00
61370e4b6c
[MATH][TOPI][NNVM] introduce trunc, round ( #1310 )
2018-06-21 13:23:54 -07:00
80e4bc029c
[FRONTEND][MXNET] Add squeeze_axis support to split operator ( #1288 )
2018-06-20 09:58:07 -07:00
42b189cbc0
[NNVM][TESTING] Add two testing symbols: dqn and dcgan ( #1294 )
2018-06-17 12:38:23 -07:00
c418e916bc
SSD support in NNVM ( #1214 )
2018-06-14 15:22:08 -07:00
76fa3ca4f6
[TOPI][IMAGE][RESIZE] Bilinear interpolation for resize and upsampling. ( #1181 )
2018-06-14 08:53:49 -07:00
fed6298b7b
[BUILD] Add clang to build matrix, -Werror ( #1273 )
2018-06-13 10:52:49 -07:00
a81ebd9040
[NNVM][FRONTEND] Tensorflow frontend support ( #1188 )
2018-06-12 22:37:35 -07:00
558cf098c0
add support for subgraphs. ( #1221 )
...
* add support for subgraphs.
* fix.
* fix.
* Fix compilation error
* Fix compilation error
* add comments.
* update comments.
* Sanity check on subgraphs when creating IndexedGraph
* avoid the overhead of sanity check.
* Stop using non-recursive DFS
* Trigger CI
* trigger CI
2018-06-12 15:46:48 -07:00
669b44c174
Fix the gemm conversion in onnx frontend ( #1241 )
2018-06-11 16:43:05 -07:00
552fa71b74
[BUILD] Upgrade build system to default python3 ( #1260 )
2018-06-11 16:36:52 -07:00
cd39697e11
Update layout.h ( #1255 )
2018-06-10 20:49:19 -07:00
4be6a30788
Unite cmake builds ( #1248 )
2018-06-08 10:23:34 -07:00
5a15664ebe
Add PReLU support to mxnet frontend ( #1249 )
2018-06-07 21:32:52 -07:00
31223fa248
[NNVM]Activations support added in Keras Frontend ( #1210 )
...
* [NNVM]Activations support added in Keras Frontend
* Helper for ELU added
* All activations test cases clubbed to one
2018-06-07 08:44:57 -07:00
34630e8147
[NNVM][FRONTEND][Keras] Support for reusing layers ( #1192 )
2018-06-06 08:44:52 -07:00
a9313787ab
[WIP] [NNVM] Fix softmax gradient ( #1201 )
...
[NNVM] Fix softmax gradient
2018-06-05 13:51:38 -07:00
ce34ae16f9
[NNVM] Elu support added in darknet frontend ( #1199 )
2018-06-04 12:12:11 -07:00
9c299a909a
leaky_relu bug fix ( #1218 )
2018-06-01 13:09:53 -07:00
59c8bc9652
[LINT] Expected, hence disable for a while. ( #1212 )
...
************* Module nnvm.top.nn
E:141, 8: Assigning to function call which doesn't return (assignment-from-no-return)
2018-05-31 10:02:35 -07:00
52f045171f
[RUNTIME] update to make runtime copy type aware
2018-05-30 22:27:44 -07:00
42608ddafb
[IO] Support cross-endian
2018-05-30 22:27:44 -07:00
55d13d47fd
Homepage URL to tvm.ai
2018-05-29 16:24:42 -07:00
5195679c5c
[DOCS] Improve docs naming, fix docs warnings
2018-05-29 08:47:00 -07:00
bb2b86201c
[DOCS] documentation merge
2018-05-29 08:47:00 -07:00
891b4e06ce
Flip operator ( #505 )
2018-05-29 08:47:00 -07:00
31edf3f7eb
Expose clip to frontend mxnet ( #512 )
2018-05-29 08:47:00 -07:00
90c1157b5f
Consider version too #508 ( #514 )
2018-05-29 08:47:00 -07:00
a9d2149711
[FRONTEND][Keras] Add test for MobileNet ( #513 )
2018-05-29 08:47:00 -07:00
c68945c582
[FRONTEND][Keras] fix reshape ( #493 )
2018-05-29 08:47:00 -07:00
39cc9c1265
fix doc ( #510 )
2018-05-29 08:47:00 -07:00
510cd5ec81
Squeeze bug fix. ( #506 )
2018-05-29 08:47:00 -07:00
69d5fcab74
Add count_include_pad support ( #498 )
...
* update tvm submodule
* Add count_include_pad support to avg_pool
2018-05-29 08:47:00 -07:00
fd6ad2742b
[FRONTEND][Keras] Fix softmax axis ( #503 )
2018-05-29 08:47:00 -07:00
8534db36b1
Increase depfiles lookup. ( #509 )
...
Not able link build/src/top/tensor/*.d.
Hence don't compile nnvm for a change in tvm/topi headers.
2018-05-29 08:47:00 -07:00
c7474c005a
Cleanup of '-Wsign-compare' warnnigs. ( #504 )
2018-05-29 08:47:00 -07:00
71ffc2b6ee
don't rename onnx input tensors ( #491 )
2018-05-29 08:47:00 -07:00
9f6e0c596c
Fixed issue #483 , removing enum dependancy ( #485 )
2018-05-29 08:47:00 -07:00
a349f5ec4f
Better to check Infer result with topi results at build time instead of leaving to a runtime error. ( #476 )
2018-05-29 08:47:00 -07:00
3c3eb3b367
add use_mt option, default to use md as mt won't work with current c++ api ( #484 )
2018-05-29 08:47:00 -07:00
7f5d22d7f6
enable AlterOpLayout to keep OP unchanged ( #471 )
2018-05-29 08:47:00 -07:00
be968fef00
[Keras] ReLU6 support ( #481 )
2018-05-29 08:47:00 -07:00
8db42c4143
add PATH to dll_path on windows ( #479 )
2018-05-29 08:47:00 -07:00
f92daf9238
use uint32_t ( #478 )
2018-05-29 08:47:00 -07:00
2864fe4e4d
dilation fixed for (1, 1) case ( #477 )
2018-05-29 08:47:00 -07:00
d47682043b
fix lint error ( #475 )
2018-05-29 08:47:00 -07:00
dd90b19eeb
add sanity check to input shape type ( #469 )
2018-05-29 08:47:00 -07:00
42d81ccaee
raise error when cannot get shape in reshape ( #467 )
...
* raise error when cannot get shape in reshape
* fix pylint
2018-05-29 08:47:00 -07:00
73775f4532
AlterOpLayout with tvm.target ( #463 )
...
* AlterOpLayout with tvm.target
* fix test
2018-05-29 08:47:00 -07:00
343eb82ca2
fix restore layout in AlterOpLayout ( #460 )
...
* fix restore layout in AlterOpLayout
* lint test case
2018-05-29 08:47:00 -07:00
1e4bb2f887
allow missing FCorrectLayout ( #457 )
...
* allow missing FCorrectLayout
* misunderstood OpMap[], fix
2018-05-29 08:47:00 -07:00
1b5877f28d
[DOCS] improve document of symbol; ( #456 )
2018-05-29 08:47:00 -07:00
11ebd8d303
support dilation in conv2d ( #439 )
2018-05-29 08:47:00 -07:00
d5744844c1
Rename FInferLayout -> FCorrectLayout ( #453 )
...
* rename FInferLayout -> FCorrectLayout
* correct stupid IDE
* update submodule tvm
2018-05-29 08:47:00 -07:00
9f8fcfc9bc
General Layout Support ( #447 )
2018-05-29 08:47:00 -07:00
fc7e9cd206
Some minor documentation issues rectified ( #450 )
2018-05-29 08:47:00 -07:00
386fbb3f43
Fix license URL ( #451 )
2018-05-29 08:47:00 -07:00
a82656d497
Support imagescalar operator for onnx ( #448 )
2018-05-29 08:47:00 -07:00
435201ee12
fix mxnet model import ( #449 )
2018-05-29 08:47:00 -07:00
4b765c511d
[OP] PReLU Support ( #394 )
2018-05-29 08:47:00 -07:00
079599c95d
Issue fix for tiny yolo ( #436 )
...
* Issue fix for tiny yolo
There was an issue in maxpool frontend for tiny-yolo. It is fixed in PR.
https://github.com/dmlc/nnvm/issues/431
* updated review comment, python 3 compatabilty
* Update darknet.py
* Updated review comment
2018-05-29 08:47:00 -07:00
d5b9f35487
Update TVM to latest ( #432 )
...
* Update TVM to latest
* remove darknet from testing due to cffi
2018-05-29 08:47:00 -07:00
7ec898d55e
[FRONTEND] DarkNet Yolo2 Frontend Support ( #377 )
2018-05-29 08:47:00 -07:00
2e836ca768
declare type name for optional<TShape> ( #429 )
...
* declare type for optional tshape
* add doc
* move code to another place
2018-05-29 08:47:00 -07:00
a53d8d0172
[PASS] Enhance scale fold axis ( #424 )
2018-05-29 08:47:00 -07:00
89c124bc89
Update tvm, fix lint due to pylint update ( #423 )
2018-05-29 08:47:00 -07:00
e722dbcbf7
Onnx opset support ( #416 )
2018-05-29 08:47:00 -07:00
f4789db696
Add optimizer ( #334 )
2018-05-29 08:47:00 -07:00
bd40bcd1a2
use caffe2.python.onnx.backend instead of onnx-caffe2 ( #418 )
2018-05-29 08:47:00 -07:00
ec5c46519b
fix onnx frontend softplus bug ( #413 )
2018-05-29 08:47:00 -07:00
8c5b4909f4
[Keras] fix dropout bug ( #399 )
2018-05-29 08:47:00 -07:00
b7b74228ff
[FIX] several bugs found when using NNVM ( #391 )
2018-05-29 08:47:00 -07:00
a523311db0
Documentation errors updated ( #392 )
2018-05-29 08:47:00 -07:00
76ca335215
fix tutorial build ( #390 )
2018-05-29 08:47:00 -07:00
6a7d1c4567
[OP] Experimental assign op ( #389 )
2018-05-29 08:47:00 -07:00
74a3f74208
Add end-to-end SGX ResNet inference example ( #388 )
2018-05-29 08:47:00 -07:00
3f22f1900a
[DOCS] add a solution to installation error because of the old submodule ( #384 )
...
* [DOCS] add a solution to installation error
* Fix the command for updating submodules
2018-05-29 08:47:00 -07:00
33ac203761
Sum operator for ONNX. ( #386 )
2018-05-29 08:47:00 -07:00
4794f3d6ec
WebGL end-to-end example ( #369 )
2018-05-29 08:47:00 -07:00
9ba95ec47a
[FRONTEND] fix same var used in single op ( #383 )
...
* fix same var used in single op
* revert to older version
2018-05-29 08:47:00 -07:00
a3a9dbeb9a
Make compiler more robust ( #378 )
2018-05-29 08:47:00 -07:00
860adec806
Initial NHWC layout support ( #376 )
...
* initial NHWC layout support
* remove layout param from softmax
* more nhwc support
* fix typo
* add nhwc layout test
* fix lint
* update tvm
* update for c++ topi
* fix lint
* update tvm
2018-05-29 08:47:00 -07:00
68c039442e
[NHWC] InferShape Layout conversion fix. ( #372 )
2018-05-29 08:47:00 -07:00
50c20b76d3
[Compilation Warning Fix]Enum constant in boolean context ( #365 )
...
The below compilation warning is fixed.
In file included from /mnt/D_DRIVE/work/nnvm_8_feb/dmlc-core/include/dmlc/any.h:16:0,
from include/nnvm/./base.h:11,
from include/nnvm/graph.h:15,
from src/compiler/fold_scale_axis.cc:6:
src/compiler/fold_scale_axis.cc: In function ‘nnvm::Graph nnvm::compiler::FoldScaleAxis(nnvm::Graph)’:
src/compiler/fold_scale_axis.cc:155:39: warning: enum constant in boolean context [-Wint-in-bool-context]
CHECK(kind == kPassTroughFirst || kMulConsumer);
^
/mnt/D_DRIVE/work/nnvm_8_feb/dmlc-core/include/dmlc/./logging.h:110:9: note: in definition of macro ‘CHECK’
if (!(x))
2018-05-29 08:47:00 -07:00
f1e0a55abc
Move dense compute back to python ( #364 )
2018-05-29 08:47:00 -07:00
5884cd01f3
Change TOPI ops to use C++ implementation where applicable ( #357 )
...
* Updated TVM version. Implemented fix for nnvm_compiler crash on exit on windows. Changed TOPI ops from using python to using C++ where applicable.
* Fix lint
* Fix lint
* Fix macro
* Fix reshape
* Update TVM to fix test fails
2018-05-29 08:47:00 -07:00
3f6423aa2d
[Compilation Warning Fix] in matrix_op.cc ( #356 )
...
The below compilation warning issue has been fixed.
src/top/tensor/matrix_op.cc:37:21: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
for (int i = 0; i < lshape.ndim() - 1; i++) oshape[i] = lshape[i];
^
src/top/tensor/matrix_op.cc:38:21: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
for (int i = 1; i < rshape.ndim(); i++) oshape[i + lshape.ndim() - 1] = rshap
2018-05-29 08:47:00 -07:00
ec37028c09
[PYLINT FIX]pylint issue fixes ( #348 )
...
Before Fix
*************
************* Module nnvm._base
R: 21, 0: Disallow trailing comma tuple (trailing-comma-tuple)
R: 27, 0: Disallow trailing comma tuple (trailing-comma-tuple)
-----------------------------------
Your code has been rated at 9.99/
After Fix
*************
siju@siju-DAEMON:~/D/Community/nnvm$ make lint
pylint python/nnvm --rcfile=/media/siju/DATA/Community/nnvm/tests/lint/pylintrc
-------------------------------------------------------------------
Your code has been rated at 10.00/10 (previous run: 9.99/10, +0.01)
2018-05-29 08:47:00 -07:00
6be636e1bb
image tranform runtime wrong on python ( #354 )
...
x = np.transpose(img, (2, 0, 1))[np.newaxis, :]
TypeError: an integer is required
2018-05-29 08:47:00 -07:00
16c28d0818
improved empty set definition ( #346 )
...
fix similar problem like https://github.com/apache/incubator-mxnet/pull/4589/files .
2018-05-29 08:47:00 -07:00
7023101178
[FIX] Fix target in tutorial ( #347 )
2018-05-29 08:47:00 -07:00
eb9442e0f2
[Convolution] Error while importing onnx model with weights. ( #345 )
...
* [Convolution] Error while imported onnx model has weights.
Fix the use_bias based on the input parameters to use bias (Ex: sqeezenet). Ref #336
* Review corrections.
2018-05-29 08:47:00 -07:00
ffcb2a5ef0
explain the lowering process in nnvm.compiler.build ( #339 )
2018-05-29 08:47:00 -07:00
6c19862176
Add CacheItem2Schedule Extension ( #338 )
...
* add CacheItem2Schedule extension
* fix lint
* move function position
* make cache item visible to frontend
2018-05-29 08:47:00 -07:00
2bb4a1e7f0
[TUTORIAL] add extern lib tutorial ( #331 )
...
* add extern lib tutorial
* do fix
2018-05-29 08:47:00 -07:00
9339d31f23
Add test for issue #327 ( #328 )
...
* Added test for adding down-/up-sampled layers
* Enabled test for adding down-/up-sampled layers
* Normalize whitespace
2018-05-29 08:47:00 -07:00
82de081cac
fix upsampling fuse pattern ( #330 )
2018-05-29 08:47:00 -07:00
9998f3302f
Fix keras frontend elementwise-ops for lists with len>2 (fixes issue #325 ) ( #326 )
...
* Added elementwise-add test
* Fix typo
* Fixed elem-wise ops for lists with len>2
2018-05-29 08:47:00 -07:00
14de1dec78
[TUTORIAL] add tutorial for mali gpu ( #313 )
...
* add tutorial for mali gpu
* update submodule version
* fix doc error
* fix server error
* fix server in test
2018-05-29 08:47:00 -07:00
963a041973
[NNVM] : Function header corrections. ( #324 )
2018-05-29 08:47:00 -07:00
5405127199
Explain how to generate module library in Quick Start tutorial ( #323 )
...
* Explain how to generate module library
* Small fix
2018-05-29 08:47:00 -07:00
f34e1744a2
[GRADIENT] Register more gradient operators ( #300 )
...
* Add conv2d max_pool backward op
* Added tests
* Fix testing
* Address comments
* Change dot to matmul
* Address comments
* Break down indicator function
* Make greater, less numpy compatible
2018-05-29 08:47:00 -07:00
4bd92a4a21
[Keras] fix convert_pooling with same pad ( #322 )
2018-05-29 08:47:00 -07:00
53361fae2a
Update frontend for keras 2.1.3 compatibility ( #314 )
...
* Keras keeps renaming properties. Update frontend for keras 2.1.3 compatibility
* Add error message when inbound_nodes is not found
2018-05-29 08:47:00 -07:00
d626eab87c
[CI] turn on keras frontend test ( #309 )
...
* [CI] turn on keras frontend test
* fix
* using tensorflow cpu version
2018-05-29 08:47:00 -07:00
2374852e6d
make shape inference of BatchNorm layout neutral ( #301 )
...
* make shape inference of BatchNorm layout neutral
* refactor to use the axis variable to do BatchNorm shape inference
* refactor to use the axis variable to do BatchNorm shape inference
* add unittest to the axis param for batch norm shape inference
2018-05-29 08:47:00 -07:00
7112dd78f0
Avoid crash when linear activation does not have alpha and beta defined ( #306 )
2018-05-29 08:47:00 -07:00
292609d885
remove dtype in model symbol ( #310 )
2018-05-29 08:47:00 -07:00
acb9fd62dd
Add support for keras upsampling ( #305 )
...
* Add support for keras upsampling
* Fix code formatting
* Fix indentation
* Fix indentation round 2
* Hide unused parameter
* Only enable UpSampling2D since the others are not currently supported by TVM
* Improve error messages and code layout
2018-05-29 08:47:00 -07:00
6a0fb6efbf
Upsampling op support ( #298 )
...
* add nnvm upsampling symbol
* add upsampling mxnet frontend
* add doc for upsampling op
* cleanup upsampling test
* minor fix
* use schedule_injective for upsampling
* upgrade tvm
2018-05-29 08:47:00 -07:00
d059dbabf0
[WIP] Onnx1.0 ( #294 )
...
* add more op for onnx 1.0
* fix syntax
* fix lint
* fix
* update 1.0
* fix
* update model
2018-05-29 08:47:00 -07:00
063cd41293
Add __declspec(dllexport) to class Node and class Op if build on Wind… ( #296 )
...
* Add __declspec(dllexport) to class Node and class Op if build on Windows, so that it can be referenced in some test code in MXNet
* correct typo
* reuse the macro in c_api.h
* revert changes on base.h
* one more place to revert
* use enclosing extern c block
* revert a change caused by copy
* remove a space
2018-05-29 08:47:00 -07:00
3dd7d6467b
dump lowered IR when debug logging ( #292 )
...
* dump lowered ir when debug log
* avoid calling tvm.lower() twice when not debug
2018-05-29 08:47:00 -07:00
8473e9face
typo in c_api.h ( #290 )
2018-05-29 08:47:00 -07:00
e06d24a68e
Add missing cmath header ( #287 )
...
f9ed337552a26cc55c7c0fb22cd54839ee13f19c added a call to std::log
without including cmath
2018-05-29 08:47:00 -07:00
Yao Wang
Pariksheet Pinjari
masahi
Lianmin Zheng
Siju
Sergey Mironov
Keren Zhou
Siva
Albin Joy
Zhi
Tang, Cheng
Tatsuya Nishiyama
Tianqi Chen
abergeron
MORITA Kazutaka
Sergei Grechanik
sgrechanik-h
Yizhi Liu
Qiao Longfei
nhynes
Dayananda V
ANSHUMAN TRIPATHY
Wenhao Hu
rajh619
Junru Shao
Thierry Moreau
Da Zheng
Tatsuya Nishiyama
Haibin Lin
larrywyang
tqchen
Ehsan M. Kermani
Tao Lv
Siva
MORITA Kazutaka
Leyuan Wang
Liangfu Chen
Tatsuya Nishiyama
Yuwei Hu
Yan Huang
Atsushi Nukariya
Zhixun Tan
Joshua Z. Zhang
alex-weaver
wl3b10s
ziheng
Yida Wang
thefiddler
Siva
yuruofeifei
guopinglong
Hector Li
Iblis Lin
Daniel M. Weeks