Lune and nihui
bedf9c6568
unaryop: extend sign, hyperbolic, expm1 and log1p support ( #6675 )
...
Add backend support for new UnaryOp operations across scalar, SIMD and Vulkan paths, including sign, sinh, cosh, asinh, acosh, atanh, expm1 and log1p.
Wire the new ops through onnx2ncnn and pnnx expression conversion, update generated Python emission, and add regression coverage for the new conversion paths.
Fix portability issues for expm1/log1p math declarations, NCNN_SIMPLEMATH builds, ONNX-enabled pnnx linking, and stabilize unary op tests across fp16, bf16 and torch exporter/version differences.
Co-authored-by: nihui <nihuini@tencent.com >
2026-05-21 16:19:58 +08:00
Evgeny Proydakov
4699b26933
Using snprintf instead of sprintf for a tools ( #6554 )
...
Signed-off-by: Evgeny Proydakov <e.proydakov@gmail.com >
2026-02-26 10:29:28 +08:00
nihui
414666a6d1
doc supported onnx operators ( #6327 )
2025-09-23 19:41:06 +08:00
nihui
a6eda68f23
disable onnx2ncnn build ( #6324 )
...
Migrate most of onnx2ncnn functions to pnnx
2025-09-23 19:06:07 +08:00
nihui
171b9d1bba
use spdx license header, copyright Tencent ( #6152 )
2025-07-08 11:01:29 +08:00
lll143653
d355b6dc5b
Add warning and recommend to use pnnx ( #5588 )
2024-07-23 16:54:35 +08:00
Amir Ramezani
7e5fa3ade3
shrink operator ( #5022 )
2023-09-20 14:08:31 +08:00
Amir Ramezani
0ea587b8c7
celu activation vulkan and onnx conversion ( #5018 )
2023-09-12 15:37:15 +08:00
Amir Ramezani
695f770eab
erf implementation ( #5012 )
...
* added erf implementation
* added testcase for erf
* added onnx2ncnn support of erf
2023-09-08 15:08:59 +08:00
nihui
cd1c0d6eab
fix build with new protobuf target ( #4955 )
2023-08-22 14:07:13 +08:00
nihui
39721eeb94
require c++17 for building with new protobuf ( #4947 )
2023-08-16 11:48:56 +08:00
huoshuai-dot
f7f6ca0033
Update onnx2ncnn.cpp ( #4754 )
2023-05-29 10:53:11 +08:00
Zhang Ge and ZhangGe6
3fce00b260
fix param parsing issue when layer/blob name exceeds 255 ( #4236 )
...
* fix param parsing issue when layer/blob name exceeds 255
* apply code-format changes
Co-authored-by: ZhangGe6 <ZhangGe6@users.noreply.github.com >
2022-10-07 20:54:14 +08:00
tpoisonooo
6fd801b6d7
feat(src/layer): add vision_transformer benchmark ( #3730 )
...
* feat(src/layer): add vision_transformer benchmark and relative layer
* refactor(testutil.h): add para for RandomMat
2022-05-17 16:13:49 +08:00
nihui
3181616439
treat old reduction axes param as failure
2022-02-06 21:20:54 +08:00
nihuini
f02b259a15
convert some pnnx reduction family
2021-12-23 16:19:25 +08:00
Jimmy
52c6502ae7
Feature/ncnntool/add install ( #3212 )
2021-09-10 14:08:10 +08:00
nihui
3d5f447260
fix onnx negative concat axis, fix #3194
2021-08-30 21:28:10 +08:00
q.yao
41aa601413
fix fuse expand ( #3189 )
2021-08-23 15:22:26 +08:00
nihuini
c5d003001c
preserve lstm hidden output
2021-07-27 19:01:41 +08:00
nihuini
9afbfdd9dd
ncnnoptimize fuse rsub rdiv with scalar
2021-07-27 17:35:30 +08:00
li mengyang
be054aacb0
some misspell ( #3046 )
...
* just a spell check
* some misspell
2021-07-03 14:47:35 +08:00
nihui
4f135e07bf
implement convolution1d and pooling1d ( #3035 )
...
* implement convolution1d and pooling1d
* add conv1d pool1d test
* fuse convolution1d activation
* update operator doc
* fix vulkan adpative pooling
2021-06-30 20:19:19 +08:00
nihuini
72282f8b1a
fix blob count when there are orphaned weight nodes, fix #2969
2021-06-01 15:23:44 +08:00
nihuini
f9e42d4ee4
drop fp16 and int8 mode from caffe2ncnn, ncnnoptimize and ncnn2int8 should be used instead
2021-04-26 14:18:57 +08:00
nihuini
6c41822d35
do not remove optional hidden output for lstm/gru/rnn
2021-03-29 19:10:22 +08:00
nihuini
5e85f447e6
convert mxnet channel padding, fix some lgtm warnings
2021-03-29 15:07:08 +08:00
nihui
d4a7abc218
fix onnx2ncnn clip without max blob, fix #2788
2021-03-26 23:58:35 +08:00
nihuini
15d63ec0f5
fuse onnx multiheadattention with same qkv blob
2021-03-15 19:08:44 +08:00
nihuini
c17eb4e208
multiheadattention layer
2021-03-05 11:45:02 +08:00
nihuini
b0d16325b1
fuse onnx binaryop with scalar
2021-03-04 11:40:29 +08:00
nihuini
f7cbcaa72b
fix onnx normalize expand ghost shape
2021-03-03 17:21:45 +08:00
nihuini
c910574b5b
fuse onnx multiheadattention
2021-03-03 16:25:33 +08:00
nihuini
f2a5ea7678
fix layernorm ghost input without affine
2021-03-02 17:32:36 +08:00
nihuini
7ac23ab34d
fuse onnx layernorm, fix 2-dim layernorm implementation, add test
2021-03-02 15:42:03 +08:00
nihuini
5b72a37544
fix megvii style shufflechannel blob count mismatch
2021-02-09 11:13:37 +08:00
nihuini
f47fbcbf83
preserve the very first onnx constant dimension size when we guess it is not the batchsize one, fix #2487
2021-02-03 15:27:45 +08:00
nihuini
0e79532975
topological sort onnx nodes, convert onnxruntime bert ops
2021-01-29 19:46:05 +08:00
nihuini
11c569fc52
fix in-between empty input node
2021-01-20 12:47:45 +08:00
nihuini
5bd07030ab
remove onnx expand shape node
2021-01-19 11:08:01 +08:00
nihuini
6572e98f66
handle onnx transpose 102 201 after lstm
2021-01-13 16:56:51 +08:00
nihui
ef3550b52f
gru and rnn layer ( #2572 )
2021-01-08 13:03:17 +08:00
nihuini
cf853bd3ce
convert onnx gru rnn graph
2021-01-07 19:26:06 +08:00
Guoxia Wang
13d0bfcfd0
fix onnx2ncnn adaptive pool conversion bug, remove no used weights node ( #2570 )
2021-01-07 18:58:24 +08:00
nihuini
8790de3507
better onnx lstm fusion
2021-01-07 18:55:46 +08:00
nihuini
9b949d65b3
fuse onnx lstm, codeformat exclude pybind11, fix #2562
2021-01-07 16:35:20 +08:00
Guoxia Wang and Restyled.io
609f63c57e
support PyTorch AdaptiveAvgPool2d and AdaptiveMaxPool2d ( #2546 )
...
* support pytorch adaptive pool
* support onnx2ncnn adaptive pool convert
* support ncnnoptimize adaptive pool param write
* fix adaptive pool out_shape order
* fix adaptive pool out_shape order, H and W can be either a int
add test case, set support_vulkan = false Pooling_vulkan::create_pipeline
* review adaptive pool
* fix typo
* add adaptive pool forward in pooling_x86.cpp pooling_arm.cpp
fix out_w, out_h id naming convention
* fix typo
* don't support packing, bf16, int8, image for adaptive pool
* Restyled by clang-format
* Restyled by astyle
* Restyled by clang-format
* Restyled by astyle
Co-authored-by: Restyled.io <commits@restyled.io >
2021-01-07 08:39:06 +08:00
nihuini
fbf0ffda53
pixelshuffle nhwc mode, convert onnx DepthToSpace mode DCR, convert mlir tf.DepthToSpace
2020-12-29 19:47:53 +08:00
nihuini
417b6874f8
fix onnx2ncnn printf format error
2020-12-29 14:11:25 +08:00
nihuini
1963672332
rewrite most onnx2ncnn node_reference and binary weight logic
2020-12-28 17:55:56 +08:00
