[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15625: Minor refactor of fully connected.
apeforest commented on a change in pull request #15625: Minor refactor of fully connected. URL: https://github.com/apache/incubator-mxnet/pull/15625#discussion_r307594376 ## File path: src/operator/nn/fully_connected-inl.h ## @@ -75,6 +75,38 @@ struct FullyConnectedParam : public dmlc::Parameter { } }; +/** + * Flatten additional dimensions after the first + * @tparam xpu + * @tparam DType + * @param tblob + * @param ctx + * @return 2 Dimensional Tensor with upper shapes collapsed + */ +template +Tensor FlattenAs2DTail(const TBlob& tblob, const OpContext& ctx) { + const TShape& shape = tblob.shape_; + Stream *stream = ctx.get_stream(); + return tblob.get_with_shape( + Shape2(shape[0], shape.ProdShape(1, shape.ndim())), stream); +} + +/** + * Flatten dimensions except last + * @tparam xpu + * @tparam DType + * @param tblob + * @param ctx + * @return 2 Dimensional tensor with front shapes collapsed + */ +template +Tensor FlattenAs2DHead(const TBlob& tblob, const OpContext& ctx) { Review comment: These two methods seem to be generic for all operators. Can we add them to a common place, also are the function names easier to understand than original code? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15625: Minor refactor of fully connected.
apeforest commented on a change in pull request #15625: Minor refactor of fully connected. URL: https://github.com/apache/incubator-mxnet/pull/15625#discussion_r307594482 ## File path: src/operator/nn/fully_connected-inl.h ## @@ -134,59 +159,52 @@ void FCBackward(const OpContext , const FullyConnectedParam , using namespace mshadow::expr; // TODO(bing): check the BLAS Handle, be careful // maybe need blas handle from context - Stream *s = ctx.get_stream(); - const mxnet::TShape& ishape = in_data[fullc::kData].shape_; - const mxnet::TShape& oshape = out_grad[fullc::kOut].shape_; - - Tensor wmat = in_data[fullc::kWeight].get(s); - Tensor data, grad, gdata; + Stream *stream = ctx.get_stream(); + Tensor wmat = in_data[fullc::kWeight].get(stream); + Tensor x, y_grad, x_grad; if (!param.flatten) { -data = in_data[fullc::kData].get_with_shape( Review comment: These variable names are used in many operators. Why rename them? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15625: Minor refactor of fully connected.
apeforest commented on a change in pull request #15625: Minor refactor of fully connected. URL: https://github.com/apache/incubator-mxnet/pull/15625#discussion_r307594376 ## File path: src/operator/nn/fully_connected-inl.h ## @@ -75,6 +75,38 @@ struct FullyConnectedParam : public dmlc::Parameter { } }; +/** + * Flatten additional dimensions after the first + * @tparam xpu + * @tparam DType + * @param tblob + * @param ctx + * @return 2 Dimensional Tensor with upper shapes collapsed + */ +template +Tensor FlattenAs2DTail(const TBlob& tblob, const OpContext& ctx) { + const TShape& shape = tblob.shape_; + Stream *stream = ctx.get_stream(); + return tblob.get_with_shape( + Shape2(shape[0], shape.ProdShape(1, shape.ndim())), stream); +} + +/** + * Flatten dimensions except last + * @tparam xpu + * @tparam DType + * @param tblob + * @param ctx + * @return 2 Dimensional tensor with front shapes collapsed + */ +template +Tensor FlattenAs2DHead(const TBlob& tblob, const OpContext& ctx) { Review comment: These two methods seem to be generic for all operators. Can we add them to a common place? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #14779: Fully connected, higher order grad
apeforest commented on a change in pull request #14779: Fully connected, higher order grad URL: https://github.com/apache/incubator-mxnet/pull/14779#discussion_r307593777 ## File path: tests/python/unittest/test_gluon.py ## @@ -21,7 +21,7 @@ import mxnet as mx from mxnet import gluon from mxnet.gluon import nn -from mxnet.test_utils import assert_almost_equal +from mxnet.test_utils import assert_almost_equal, same Review comment: But there is no change in the test in this file. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #14779: Fully connected, higher order grad
apeforest commented on a change in pull request #14779: Fully connected, higher order grad URL: https://github.com/apache/incubator-mxnet/pull/14779#discussion_r307593378 ## File path: src/operator/nn/fully_connected-inl.h ## @@ -249,6 +285,114 @@ void FullyConnectedGradCompute(const nnvm::NodeAttrs& attrs, } } + + +/// +// Inputs are: +// o_x_grad : head gradient for x_grad +// o_w_grad : head gradient for w_grad +// o_b_grad : if param.no_bias is false +// o_y : head gradient of y +// +// outputs are: +// o_y_grad : gradient of o_y +// x_grad_grad : o_y * o_w_grad +// w_grad_grad : o_y.T * o_x_grad +// b_grad_grad: if param.no_bias is false +// +// For implementation details see this PR: https://github.com/apache/incubator-mxnet/pull/14779 + +/** + * Second order gradient for Fully Connected + * x_grad_grad = o_y * o_w_grad + * w_grad_grad = o_y.T * o_x_grad + * + * @tparam xpu + * @tparam DType + * @param attrs + * @param ctx + * @param inputs + * @param req + * @param outputs + */ +template +void FullyConnectedGradGradCompute(const nnvm::NodeAttrs& attrs, + const OpContext& ctx, + const std::vector& inputs, + const std::vector& req, + const std::vector& outputs) { + using namespace std; + using namespace fullc; + Stream *stream = ctx.get_stream(); + const FullyConnectedParam& param = nnvm::get(attrs.parsed); + const size_t num_inputs = param.no_bias ? 3U : 4U; + // outputs are: o_x_grad, o_w_grad, o_y || o_x_grad, o_w_grad, o_b_grad, o_y + const size_t num_outputs = 3U; + CHECK_EQ(inputs.size(), num_inputs); + CHECK_EQ(outputs.size(), num_outputs); + CHECK_EQ(req.size(), num_outputs); + + // inputs + Tensor o_x_grad; + Tensor o_w_grad; + Tensor o_y; + // unused + // Tensor o_b_grad; + + // outputs + Tensor o_y_grad; + TBlob o_y_grad_blob = outputs[kOyGrad]; + Tensor x_grad_grad; + Tensor w_grad_grad; + Tensor b_grad_grad; + size_t o_y_idx = std::numeric_limits::max(); + if (param.no_bias) +o_y_idx = kOy; + else +o_y_idx = kOyBias; + if (!param.flatten) { +o_x_grad = FlattenAs2DHead(inputs[kOxGrad], ctx); +o_w_grad = inputs[kOwGrad].get(stream); +o_y = FlattenAs2DHead(inputs[o_y_idx], ctx); +x_grad_grad = FlattenAs2DHead(outputs[kXGradGrad], ctx); +w_grad_grad = FlattenAs2DHead(outputs[kWGradGrad], ctx); + } else { +o_x_grad = FlattenAs2DTail(inputs[kOxGrad], ctx); +o_w_grad = FlattenAs2DTail(inputs[kOwGrad], ctx); +o_y = inputs[o_y_idx].get(stream); +x_grad_grad = FlattenAs2DTail(outputs[kXGradGrad], ctx); +w_grad_grad = FlattenAs2DTail(outputs[kWGradGrad], ctx); + } + linalg_gemm(o_y, o_w_grad, x_grad_grad, false, false, stream); + linalg_gemm(o_y, o_x_grad, w_grad_grad, true, false, stream); + // 3rd order not supported + Fill(stream, o_y_grad_blob, kWriteTo, static_cast(0)); + /* TODO(larroy) bias is not supported yet as there's no bias input to backward. Bias grad grad is Review comment: @sxjscience Could you please review if this is correct? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #14779: Fully connected, higher order grad
apeforest commented on a change in pull request #14779: Fully connected, higher order grad URL: https://github.com/apache/incubator-mxnet/pull/14779#discussion_r307592893 ## File path: src/operator/nn/fully_connected-inl.h ## @@ -47,7 +48,24 @@ namespace fullc { enum FullyConnectedOpInputs {kData, kWeight, kBias}; enum FullyConnectedOpResource {kTempSpace}; enum FullyConnectedOpOutputs {kOut}; -} // fullc +enum FullyConnectedGradGradOutputs { Review comment: nit: Can we keep the same code style as previous enums This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #14779: Fully connected, higher order grad
apeforest commented on a change in pull request #14779: Fully connected, higher order grad URL: https://github.com/apache/incubator-mxnet/pull/14779#discussion_r307592850 ## File path: src/operator/nn/fully_connected-inl.h ## @@ -47,7 +48,24 @@ namespace fullc { enum FullyConnectedOpInputs {kData, kWeight, kBias}; enum FullyConnectedOpResource {kTempSpace}; enum FullyConnectedOpOutputs {kOut}; -} // fullc +enum FullyConnectedGradGradOutputs { + kOyGrad, + kXGradGrad, + kWGradGrad, + kBGradGrad +}; +enum Inputs { Review comment: A more specific name? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] iblis17 commented on issue #15609: update julia install doc
iblis17 commented on issue #15609: update julia install doc URL: https://github.com/apache/incubator-mxnet/pull/15609#issuecomment-515312374 looks fine. Could you rebase to re-trigger CI again? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] iblis17 commented on issue #15609: update julia install doc
iblis17 commented on issue #15609: update julia install doc URL: https://github.com/apache/incubator-mxnet/pull/15609#issuecomment-515312328 http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-15609/4/install/index.html?platform=Linux=Julia=CPU This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] wuxun-zhang commented on issue #15659: Dropout produces wrong mask with MKL-DNN
wuxun-zhang commented on issue #15659: Dropout produces wrong mask with MKL-DNN URL: https://github.com/apache/incubator-mxnet/issues/15659#issuecomment-515312297 OK. I'll look into this issue. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15285: Graph dumper
apeforest commented on a change in pull request #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#discussion_r307590972 ## File path: src/imperative/imperative.cc ## @@ -353,6 +356,11 @@ std::vector Imperative::Backward( << "There are no inputs in computation graph that require gradients."; } + if (backward_graph_dump_enabled_) { +std::cout << "Forward graph: " << std::endl; Review comment: Can we leave this to the developer? I think we can just have a function DebugGraph() and not putting all the std::cout logic in a function block. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15285: Graph dumper
apeforest commented on a change in pull request #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#discussion_r307590664 ## File path: include/mxnet/imperative.h ## @@ -191,6 +191,7 @@ class Imperative { std::atomic variable_count_{0}; /*! \brief default backward bulk size */ int backward_bulk_size_{0}; + bool backward_graph_dump_enabled_{}; Review comment: not needed. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] apeforest commented on a change in pull request #15285: Graph dumper
apeforest commented on a change in pull request #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#discussion_r307590642 ## File path: include/mxnet/imperative.h ## @@ -160,8 +160,8 @@ class Imperative { private: friend class NDArray; - /*! \brief make constructor protected. */ - Imperative() { + Imperative() : +backward_graph_dump_enabled_(dmlc::GetEnv("MXNET_BACKWARD_GRAPH_DUMP", false)) { Review comment: Why do we need a class member for a debugging utility? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] kshitij12345 commented on a change in pull request #15109: [DOC] refine autograd docs
kshitij12345 commented on a change in pull request #15109: [DOC] refine autograd docs URL: https://github.com/apache/incubator-mxnet/pull/15109#discussion_r307589169 ## File path: docs/api/python/autograd/autograd.md ## @@ -76,7 +82,63 @@ Detailed tutorials are available in Part 1 of [the MXNet gluon book](http://gluon.mxnet.io/). +# Higher order gradient + +Some operators support higher order gradients. Meaning that you calculate the gradient of the +gradient. For this the operator's backward must be as well differentiable. Some operators support +differentiating multiple times, and others two, most just once. + +For calculating higher order gradients, we can use the `mx.autograd.grad` function while recording +and then call backward, or call `mx.autograd.grad` two times. If we do the later is important that +the first call uses `create_graph=True` and `retain_graph=True` and the second call uses +`create_graph=False` and `retain_graph=True`. Otherwise we will not get the results that we want. If Review comment: So I guess `retain` keeps the intermediate gradient buffer but not the node and `create_graph` keeps the nodes as well? Thanks for the pointer. Will try to look into the imperative.cc. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zixuanweeei commented on a change in pull request #15621: [WIP] MKL-DNN LBR-GRU Inference Integration (FP32 LBR-GRU)
zixuanweeei commented on a change in pull request #15621: [WIP] MKL-DNN LBR-GRU Inference Integration (FP32 LBR-GRU) URL: https://github.com/apache/incubator-mxnet/pull/15621#discussion_r307588514 ## File path: src/operator/nn/mkldnn/mkldnn_rnn_impl.h ## @@ -118,7 +151,7 @@ static size_t GetMKLDNNRNNCacheMemorySize(int L, break; case rnn_enum::kGru: size = 2 * (D * (I + H) * 3 * H + (L - 1) * D * (D * H + H) * 3 * H + - L * D * 2 * N * H) + T * N * D * H + L * 2 * D * 3 * H + (L + 2) * D * 2 * N * H + + L * D * 2 * N * H) + T * N * D * H + L * 2 * D * 4 * H + (L + 2) * D * 2 * N * H + Review comment: @marcoabreu Sorry for the late update. I have renamed this part. Would you mind checking it again? Thanks. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zixuanweeei commented on issue #15631: [Flaky test] Skip test_operator_gpu.test_convolution_independent_gradients
zixuanweeei commented on issue #15631: [Flaky test] Skip test_operator_gpu.test_convolution_independent_gradients URL: https://github.com/apache/incubator-mxnet/pull/15631#issuecomment-515308138 @TaoLv Please take some reviews on this PR again. Thanks. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] TaoLv commented on issue #15659: Dropout produces wrong mask with MKL-DNN
TaoLv commented on issue #15659: Dropout produces wrong mask with MKL-DNN URL: https://github.com/apache/incubator-mxnet/issues/15659#issuecomment-515307535 Some background: MKL-DNN itself doesn't support dropout primitive. When MXNet is built with MKL-DNN, it will use functions from VSL to implement dropout. It's kind of legacy code from MKL2017 integration. That's why it cannot be disabled by `MXNET_MKLDNN_ENABLED=0`. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] pengzhao-intel commented on issue #15518: [MKLDNN] Enable subgraph backend mkldnn by default.
pengzhao-intel commented on issue #15518: [MKLDNN] Enable subgraph backend mkldnn by default. URL: https://github.com/apache/incubator-mxnet/pull/15518#issuecomment-515297073 Merging now. @MXNet/community Feel free to ping me or @ZhennanQin if there are issues after the PR merging since the default behaviors are changed. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch master updated (b00bb81 -> e98fea3)
This is an automated email from the ASF dual-hosted git repository. patriczhao pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git. from b00bb81 [Opperf] Add array rearrange operators to opperf (#15606) add e98fea3 [MKLDNN] Enable subgraph backend mkldnn by default. (#15518) No new revisions were added by this update. Summary of changes: cpp-package/example/inference/README.md| 3 - docs/faq/env_var.md| 3 +- docs/tutorials/c++/subgraphAPI.md | 11 +- docs/tutorials/mkldnn/MKLDNN_README.md | 18 +- example/quantization/README.md | 29 +- example/ssd/README.md | 2 - .../scala/org/apache/mxnet/OperatorSuite.scala | 8 +- src/c_api/c_api_symbolic.cc| 8 +- src/c_api/c_api_test.cc| 8 +- src/executor/graph_executor.cc | 399 + src/operator/subgraph/build_subgraph.cc| 25 +- src/operator/subgraph/default_subgraph_property.cc | 1 + .../subgraph/default_subgraph_property_v2.cc | 2 + .../subgraph/mkldnn/mkldnn_conv_property.h | 14 +- .../subgraph/mkldnn/mkldnn_subgraph_property.cc| 8 + src/operator/subgraph/subgraph_property.h | 115 +- src/operator/subgraph/tensorrt/tensorrt.cc | 2 + tests/python/mkl/test_quantization_mkldnn.py | 4 + tests/python/unittest/test_operator.py | 135 +++ 19 files changed, 504 insertions(+), 291 deletions(-)
[GitHub] [incubator-mxnet] pengzhao-intel commented on issue #15659: Dropout produces wrong mask with MKL-DNN
pengzhao-intel commented on issue #15659: Dropout produces wrong mask with MKL-DNN URL: https://github.com/apache/incubator-mxnet/issues/15659#issuecomment-515297883 @wuxun-zhang please take a look for this issue. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on issue #15285: Graph dumper
larroy commented on issue #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#issuecomment-515280608 @anirudh2290 @ptrendx @szha please review again, I removed the offending code. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy edited a comment on issue #15285: Graph dumper
larroy edited a comment on issue #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#issuecomment-515280608 @anirudh2290 @ptrendx @szha please review again, I removed the uber engineered code. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ngunauj commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
ngunauj commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656#issuecomment-515278914 > The documentation seems to be incomplete. > https://mxnet.incubator.apache.org/api/python/ndarray/ndarray.html#mxnet.ndarray.UpSampling > > It doesn't mention about weight when using `bilinear`. > Digging into the code, > https://github.com/apache/incubator-mxnet/blob/8158ba4b0f1ebd696ec09a0b1aa6031bacb60740/src/operator/nn/upsampling.cc#L123-L173 > > It mentions about requiring the weight and also gives an example. > I have verified that the example works, > > ``` > from mxnet import nd > x = nd.ones((1,3,3,3)) # NCHW > w = nd.ones((3,1,4,4)) > upsampled = nd.UpSampling(x, w, scale=2, sample_type='bilinear', num_filter=3) > ``` ok i will try..try Thanks This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ngunauj commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
ngunauj commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656#issuecomment-515279181 > The documentation seems to be incomplete. > https://mxnet.incubator.apache.org/api/python/ndarray/ndarray.html#mxnet.ndarray.UpSampling > > It doesn't mention about weight when using `bilinear`. > Digging into the code, > https://github.com/apache/incubator-mxnet/blob/8158ba4b0f1ebd696ec09a0b1aa6031bacb60740/src/operator/nn/upsampling.cc#L123-L173 > > It mentions about requiring the weight and also gives an example. > I have verified that the example works, > > ``` > from mxnet import nd > x = nd.ones((1,3,3,3)) # NCHW > w = nd.ones((3,1,4,4)) > upsampled = nd.UpSampling(x, w, scale=2, sample_type='bilinear', num_filter=3) > ``` ok i will try..try Thanks @kshitij12345 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ngunauj removed a comment on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
ngunauj removed a comment on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656#issuecomment-515278914 > The documentation seems to be incomplete. > https://mxnet.incubator.apache.org/api/python/ndarray/ndarray.html#mxnet.ndarray.UpSampling > > It doesn't mention about weight when using `bilinear`. > Digging into the code, > https://github.com/apache/incubator-mxnet/blob/8158ba4b0f1ebd696ec09a0b1aa6031bacb60740/src/operator/nn/upsampling.cc#L123-L173 > > It mentions about requiring the weight and also gives an example. > I have verified that the example works, > > ``` > from mxnet import nd > x = nd.ones((1,3,3,3)) # NCHW > w = nd.ones((3,1,4,4)) > upsampled = nd.UpSampling(x, w, scale=2, sample_type='bilinear', num_filter=3) > ``` ok i will try..try Thanks This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15109: [DOC] refine autograd docs
larroy commented on a change in pull request #15109: [DOC] refine autograd docs URL: https://github.com/apache/incubator-mxnet/pull/15109#discussion_r307563953 ## File path: docs/api/python/autograd/autograd.md ## @@ -76,7 +82,63 @@ Detailed tutorials are available in Part 1 of [the MXNet gluon book](http://gluon.mxnet.io/). +# Higher order gradient + +Some operators support higher order gradients. Meaning that you calculate the gradient of the +gradient. For this the operator's backward must be as well differentiable. Some operators support +differentiating multiple times, and others two, most just once. + +For calculating higher order gradients, we can use the `mx.autograd.grad` function while recording +and then call backward, or call `mx.autograd.grad` two times. If we do the later is important that +the first call uses `create_graph=True` and `retain_graph=True` and the second call uses +`create_graph=False` and `retain_graph=True`. Otherwise we will not get the results that we want. If Review comment: @apeforest @kshitij12345 I don't think pasting the graph here helps because it doesn't change, it's only related to attached autograd to arrays from my point of view. There's a difference, you can retain and not create if you want to calculate gradients again. If you reuse the outputs in the graph you should create. The details are in imperative.cc Backward, but they have in fact slightly different behaviour. You can try yourself, but if you use heads=x_grad_grad in the third gradient, you need to have create=True in the previous call. ![graphviz (4)](https://user-images.githubusercontent.com/928489/61920546-ccafff80-af0e-11e9-8735-259a2024de2a.png) ``` with autograd.record(): y = op(x) x_grad = autograd.grad(heads=y, variables=x, head_grads=y_grad, create_graph=True, retain_graph=True)[0] x_grad_grad = autograd.grad(heads=x_grad, variables=x, head_grads=head_grad_grads, create_graph=False, retain_graph=True)[0] x_grad_grad_grad = autograd.grad(heads=x_grad, variables=x, head_grads=head_grad_grads, create_graph=False, retain_graph=True)[0] #x_grad.backward(head_grad_grads) ``` In any case I don't see this is a reason to block the PR. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] mseth10 commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
mseth10 commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307560868 ## File path: tests/python/unittest/test_library_loading.py ## @@ -0,0 +1,29 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +import mxnet as mx +import os +from mxnet.test_utils import download + +def test_library_loading(): +if (os.name=='posix'): +lib = 'mylib.so' +elif (os.name=='nt'): +lib = 'mylib.dll' + +fname = mx.test_utils.download('https://mxnet-demo-models.s3.amazonaws.com/lib_binary/'+lib) Review comment: Added. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] mseth10 commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
mseth10 commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307560982 ## File path: src/common/library.cc ## @@ -0,0 +1,89 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +/*! + * Copyright (c) 2015 by Contributors + * \file library.cc + * \brief Dynamically loading accelerator library + * and accessing its functions + */ + +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) +#include +#else +#include +#endif + +#include "library.h" + + +/* +Loads the dynamic shared library file +Parameter: Library file location +Returns: handle for the loaded library, NULL if loading unsuccessful +*/ +void* load_lib(const char* path) { + void *handle; +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) + handle = LoadLibrary(path); + if (!handle) { +char *lpMsgBuf; +unsigned long dw = GetLastError(); +FormatMessage( +FORMAT_MESSAGE_ALLOCATE_BUFFER | +FORMAT_MESSAGE_FROM_SYSTEM | +FORMAT_MESSAGE_IGNORE_INSERTS, +NULL, +dw, +MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), +lpMsgBuf, Review comment: Done. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #15662: cudnn Dropout reproducibility
zachgk commented on issue #15662: cudnn Dropout reproducibility URL: https://github.com/apache/incubator-mxnet/issues/15662#issuecomment-515275803 I did a quick recursive grep and found several other usages of the rand function. Here is the full list I found (some of these may be correct): - https://github.com/apache/incubator-mxnet/blob/master/src/operator/nn/dropout-inl.h#L95 - https://github.com/apache/incubator-mxnet/blob/master/src/operator/nn/dropout-inl.h#L491 - https://github.com/apache/incubator-mxnet/blob/master/src/operator/random/sampler.h#L110 - https://github.com/apache/incubator-mxnet/blob/master/src/operator/rnn-inl.h#L1551 - https://github.com/apache/incubator-mxnet/blob/master/src/operator/rnn_impl.h#L159 - https://github.com/apache/incubator-mxnet/blob/master/src/operator/rnn_impl.h#L1004 - https://github.com/apache/incubator-mxnet/blob/master/src/operator/rnn_impl.h#L1891 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] mseth10 commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
mseth10 commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307560428 ## File path: python/mxnet/library.py ## @@ -0,0 +1,51 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +# coding: utf-8 +"""Library management API of mxnet.""" +from __future__ import absolute_import +import ctypes +import os +from .base import _LIB +from .base import check_call + +def load(path): +"""Loads library dynamically. + +Parameters +- +path : Path to library .so file + +Returns +- +void +""" +#check if path exists +if not os.path.exists(path): Review comment: Added this check. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy removed a comment on issue #15285: Graph dumper
larroy removed a comment on issue #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#issuecomment-515257152 It's ironic, the amount of questionable code that has gone in the repo, and having a single header data structure with tests is a big issue. No worries, I will remove it. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy removed a comment on issue #15285: Graph dumper
larroy removed a comment on issue #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#issuecomment-515269966 Just curious, am I the only one who sees the design problems of the indexed graph class? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. marcoabreu pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 11f1c0c Bump the publish timestamp. 11f1c0c is described below commit 11f1c0c009c14acc11f1de5560982b7eb5a34064 Author: mxnet-ci AuthorDate: Fri Jul 26 01:17:40 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..a7cd79a --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Fri Jul 26 01:17:40 UTC 2019
[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15662: cudnn Dropout reproducibility
mxnet-label-bot commented on issue #15662: cudnn Dropout reproducibility URL: https://github.com/apache/incubator-mxnet/issues/15662#issuecomment-515271817 Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended labels: Cuda, Bug This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] eric-haibin-lin opened a new issue #15662: cudnn Dropout reproducibility
eric-haibin-lin opened a new issue #15662: cudnn Dropout reproducibility URL: https://github.com/apache/incubator-mxnet/issues/15662 https://github.com/apache/incubator-mxnet/blob/master/src/operator/nn/dropout-inl.h#L491-L493 The seed of CUDNN dropout does not respect the random seed in mxnet This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on issue #15285: Graph dumper
larroy commented on issue #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#issuecomment-515269966 Just curious, am I the only one who sees the design problems of the indexed graph class? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15331: [fix] missing input log higher order.
larroy commented on a change in pull request #15331: [fix] missing input log higher order. URL: https://github.com/apache/incubator-mxnet/pull/15331#discussion_r307541582 ## File path: src/operator/tensor/elemwise_unary_op_basic.cc ## @@ -1117,15 +1117,15 @@ MXNET_OPERATOR_REGISTER_BINARY_WITH_SPARSE_CPU_DR(_backward_log10, unary_bwd) .set_attr("FGradient", [](const nnvm::NodePtr& n, const std::vector& ograds) { -// ograds[0]: dL/dxgrad +// ograds[0]: dL/dygrad Review comment: I think the notation is complicating us in excess as it gets pretty hairy. It's the head gradient of the previous (output) node, which has shape of x, and x_grad. So it has to be related to x, not y. I think in Lagrange notation it would be $$F_{L_x}$$ (derivative of some head function with respect to the derivative of the first loss wrt to x. (x_grad). This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] liav-teichner commented on issue #15486: mxnet profiler yields exception when multithreaded mxndarray IO operations occur in the background
liav-teichner commented on issue #15486: mxnet profiler yields exception when multithreaded mxndarray IO operations occur in the background URL: https://github.com/apache/incubator-mxnet/issues/15486#issuecomment-515259583 Hi, I am a co-worker of Aviv and would like to ask a further question: while we target moving to Gluon API and Gluon data loading, our current code uses Module API and DataIter classes. To achieve acceptable data loading performance we spawn child processes that load data from the disk, augment the data and use NDArray.save to save it to shared memory. The main process that executes the training loop has a background thread that continuously calls a single MXNet function - NDArray.load to load NDArray from shared memory. Will this result in unexpected behaviour? (I.e. having a single background thread that calld NDArray.load dyring training). If it does, what's the easiest way to replace it with code that won't result in unexpected behaviour? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on issue #15285: Graph dumper
larroy commented on issue #15285: Graph dumper URL: https://github.com/apache/incubator-mxnet/pull/15285#issuecomment-515257152 It's ironic, the amount of questionable code that has gone in the repo, and having a single header data structure with tests is a big issue. No worries, I will remove it. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15331: [fix] missing input log higher order.
larroy commented on a change in pull request #15331: [fix] missing input log higher order. URL: https://github.com/apache/incubator-mxnet/pull/15331#discussion_r307541582 ## File path: src/operator/tensor/elemwise_unary_op_basic.cc ## @@ -1117,15 +1117,15 @@ MXNET_OPERATOR_REGISTER_BINARY_WITH_SPARSE_CPU_DR(_backward_log10, unary_bwd) .set_attr("FGradient", [](const nnvm::NodePtr& n, const std::vector& ograds) { -// ograds[0]: dL/dxgrad +// ograds[0]: dL/dygrad Review comment: I think the notation is complicating us in excess as it gets pretty hairy. It's the head gradient of the previous (output) node, which has shape of x, and x_grad. So it has to be related to x, not y. I think in Lagrange notation it would be $$F_{L_x}$$ This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15331: [fix] missing input log higher order.
larroy commented on a change in pull request #15331: [fix] missing input log higher order. URL: https://github.com/apache/incubator-mxnet/pull/15331#discussion_r307541582 ## File path: src/operator/tensor/elemwise_unary_op_basic.cc ## @@ -1117,15 +1117,15 @@ MXNET_OPERATOR_REGISTER_BINARY_WITH_SPARSE_CPU_DR(_backward_log10, unary_bwd) .set_attr("FGradient", [](const nnvm::NodePtr& n, const std::vector& ograds) { -// ograds[0]: dL/dxgrad +// ograds[0]: dL/dygrad Review comment: I think the notation is complicating us in excess as it gets pretty hairy. It's the head gradient of the previous (output) node, which has shape of x, and x_grad. So it has to be related to x, not y. I think in Lagrange notation it would be $F_{L_x}$ This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larroy commented on a change in pull request #15331: [fix] missing input log higher order.
larroy commented on a change in pull request #15331: [fix] missing input log higher order. URL: https://github.com/apache/incubator-mxnet/pull/15331#discussion_r307541582 ## File path: src/operator/tensor/elemwise_unary_op_basic.cc ## @@ -1117,15 +1117,15 @@ MXNET_OPERATOR_REGISTER_BINARY_WITH_SPARSE_CPU_DR(_backward_log10, unary_bwd) .set_attr("FGradient", [](const nnvm::NodePtr& n, const std::vector& ograds) { -// ograds[0]: dL/dxgrad +// ograds[0]: dL/dygrad Review comment: I think the notation is complicating us in excess as it gets pretty hairy. It's the head gradient of the previous (output) node, which has shape of x, and x_grad. So it has to be related to x, not y. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307533906 ## File path: src/common/library.cc ## @@ -0,0 +1,89 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +/*! + * Copyright (c) 2015 by Contributors + * \file library.cc + * \brief Dynamically loading accelerator library + * and accessing its functions + */ + +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) +#include +#else +#include +#endif + +#include "library.h" + + +/* +Loads the dynamic shared library file +Parameter: Library file location +Returns: handle for the loaded library, NULL if loading unsuccessful +*/ +void* load_lib(const char* path) { + void *handle; +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) + handle = LoadLibrary(path); + if (!handle) { +char *lpMsgBuf; +unsigned long dw = GetLastError(); +FormatMessage( +FORMAT_MESSAGE_ALLOCATE_BUFFER | +FORMAT_MESSAGE_FROM_SYSTEM | +FORMAT_MESSAGE_IGNORE_INSERTS, +NULL, +dw, +MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), +lpMsgBuf, Review comment: we should follow the MS example and pass casting it to (LPSTR) ie. (char*) https://docs.microsoft.com/en-us/windows/win32/debug/retrieving-the-last-error-code This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307533543 ## File path: tests/python/unittest/test_library_loading.py ## @@ -0,0 +1,29 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +import mxnet as mx +import os +from mxnet.test_utils import download + +def test_library_loading(): +if (os.name=='posix'): +lib = 'mylib.so' +elif (os.name=='nt'): +lib = 'mylib.dll' + +fname = mx.test_utils.download('https://mxnet-demo-models.s3.amazonaws.com/lib_binary/'+lib) Review comment: lets add this code after line 28: ``` fname = os.path.abspath(fname) ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307533408 ## File path: python/mxnet/library.py ## @@ -0,0 +1,51 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +# coding: utf-8 +"""Library management API of mxnet.""" +from __future__ import absolute_import +import ctypes +import os +from .base import _LIB +from .base import check_call + +def load(path): +"""Loads library dynamically. + +Parameters +- +path : Path to library .so file + +Returns +- +void +""" +#check if path exists +if not os.path.exists(path): Review comment: lets also require users to path an absolute path, so that we dont have to worry about when a user gives a path and the dynamic loader cannot find it. If we force users to provide an absolute path, it should always be found by the dynamic loader when we call dlopen. we can check by doing: ``` if not os.path.isabs(path): print('load path "%s" is not an absolute path' % path) return ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya edited a comment on issue #15593: Large Index Support for Slice
ChaiBapchya edited a comment on issue #15593: Large Index Support for Slice URL: https://github.com/apache/incubator-mxnet/pull/15593#issuecomment-515239154 How about using `repeat` operator for converting any dimension array to 1D? ``` >> mx.nd.repeat(mx.nd.array([[[1,2],[3,4]]]),repeats=1) [1. 2. 3. 4.] ``` ``` >>> mx.nd.repeat(mx.nd.array([[[1,2],[3,4]],[[5,6],[7,8]]]),repeats=1) [1. 2. 3. 4. 5. 6. 7. 8.] ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #15593: Large Index Support for Slice
ChaiBapchya commented on issue #15593: Large Index Support for Slice URL: https://github.com/apache/incubator-mxnet/pull/15593#issuecomment-515239154 How about using `repeat` operator for converting any dimension array to 1D? ``` >> mx.nd.repeat(mx.nd.array([[[1,2],[3,4]]]),repeats=1) [1. 2. 3. 4.] ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] access2rohit commented on issue #15593: Large Index Support for Slice
access2rohit commented on issue #15593: Large Index Support for Slice URL: https://github.com/apache/incubator-mxnet/pull/15593#issuecomment-515236970 @apeforest : MXNet NDArray flatten still flattens to 2d array instead of 1D array. Can you suggest which operator to use in order to flatten to 1D ? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya opened a new pull request #15661: [OpPerf] PDF Random ops break the utility
ChaiBapchya opened a new pull request #15661: [OpPerf] PDF Random ops break the utility URL: https://github.com/apache/incubator-mxnet/pull/15661 ## Description ## Newly add PDF Random ops - https://github.com/apache/incubator-mxnet/pull/14617 Caused the opperf.py to error out due to parameter issues. Hence added new param `sample` to handle that ## Checklist ## ### Essentials ### Please feel free to remove inapplicable items for your PR. - [x] Changes are complete (i.e. I finished coding on this PR) - [x] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ### Changes ### - [ ] added param ## Comments ## Ran `python benchmark/opperf/opperf.py` successfully without error on CPU & GPU This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet] branch master updated: [Opperf] Add array rearrange operators to opperf (#15606)
This is an automated email from the ASF dual-hosted git repository. skm pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git The following commit(s) were added to refs/heads/master by this push: new b00bb81 [Opperf] Add array rearrange operators to opperf (#15606) b00bb81 is described below commit b00bb81fd4346196d7e059e3b051d1c74ba6e572 Author: Chaitanya Prakash Bapat AuthorDate: Thu Jul 25 15:12:42 2019 -0700 [Opperf] Add array rearrange operators to opperf (#15606) * add array rearrange operators to opperf * Trigger notification * 4d tensor, param support * new line * add alias logic --- benchmark/opperf/nd_operations/README.md | 5 -- benchmark/opperf/nd_operations/array_rearrange.py | 57 ++ benchmark/opperf/nd_operations/binary_operators.py | 4 +- benchmark/opperf/opperf.py | 4 ++ benchmark/opperf/rules/default_params.py | 15 +- benchmark/opperf/utils/op_registry_utils.py| 22 + benchmark/opperf/utils/profiler_utils.py | 20 ++-- 7 files changed, 116 insertions(+), 11 deletions(-) diff --git a/benchmark/opperf/nd_operations/README.md b/benchmark/opperf/nd_operations/README.md index b98a0d3..321158c 100644 --- a/benchmark/opperf/nd_operations/README.md +++ b/benchmark/opperf/nd_operations/README.md @@ -66,9 +66,7 @@ 46. linalg_extractdiag 47. sgd_mom_update 48. SequenceLast -50. flip 51. SequenceReverse -52. swapaxes 53. SVMOutput 54. linalg_trsm 55. where @@ -82,7 +80,6 @@ 63. mp_sgd_mom_update 64. choose_element_0index 65. tile -66. space_to_depth 67. gather_nd 69. SequenceMask 70. reshape_like @@ -110,14 +107,12 @@ 94. broadcast_like 95. Embedding 96. linalg_makediag -97. transpose 98. linalg_syrk 99. squeeze 101. ROIPooling 102. ftrl_update 103. SliceChannel 104. slice_like -105. depth_to_space 106. linalg_maketrian 108. pad 109. LayerNorm diff --git a/benchmark/opperf/nd_operations/array_rearrange.py b/benchmark/opperf/nd_operations/array_rearrange.py new file mode 100644 index 000..151127c --- /dev/null +++ b/benchmark/opperf/nd_operations/array_rearrange.py @@ -0,0 +1,57 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +import mxnet as mx +from benchmark.opperf.utils.benchmark_utils import run_op_benchmarks +from benchmark.opperf.utils.op_registry_utils import get_all_rearrange_operators + +"""Performance benchmark tests for MXNet NDArray Rearrange Operators. + +1. transpose +2. swapaxes +3. flip +4. depth_to_space +5. space_to_depth +""" + + +def run_rearrange_operators_benchmarks(ctx=mx.cpu(), dtype='float32', warmup=25, runs=100): +"""Runs benchmarks with the given context and precision (dtype) for all the +rearrange operators in MXNet. + +Parameters +-- +ctx: mx.ctx +Context to run benchmarks +dtype: str, default 'float32' +Precision to use for benchmarks +warmup: int, default 25 +Number of times to run for warmup +runs: int, default 100 +Number of runs to capture benchmark results + +Returns +--- +Dictionary of results. Key -> Name of the operator, Value -> Benchmark results. + +""" +# Fetch all optimizer operators +mx_rearrange_ops = get_all_rearrange_operators() + +# Run benchmarks +mx_rearrange_op_results = run_op_benchmarks(mx_rearrange_ops, dtype, ctx, warmup, runs) +return mx_rearrange_op_results diff --git a/benchmark/opperf/nd_operations/binary_operators.py b/benchmark/opperf/nd_operations/binary_operators.py index cca8f9d..1898f5d 100644 --- a/benchmark/opperf/nd_operations/binary_operators.py +++ b/benchmark/opperf/nd_operations/binary_operators.py @@ -39,7 +39,7 @@ from benchmark.opperf.utils.op_registry_utils import get_all_broadcast_binary_op def run_mx_binary_broadcast_operators_benchmarks(ctx=mx.cpu(), dtype='float32', warmup=25, runs=100): -"""Runs benchmarks with the given context and precision (dtype)for all the binary +"""Runs benchmarks with the given context and precision (dtype) for all the binary broadcast operators in MXNet.
[GitHub] [incubator-mxnet] sandeep-krishnamurthy merged pull request #15606: [Opperf] Add array rearrange operators to opperf
sandeep-krishnamurthy merged pull request #15606: [Opperf] Add array rearrange operators to opperf URL: https://github.com/apache/incubator-mxnet/pull/15606 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 edited a comment on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0)
Zha0q1 edited a comment on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0) URL: https://github.com/apache/incubator-mxnet/issues/9686#issuecomment-515231825 I would like to bring up one issue with profiler: currently, there is a flag `kSimbolic` that supposedly should control whether to profiler operators called in symbolic mode. However, there is rarely a use case where users would want to set it to `False`; also, in the backend, this flag is not used at all, meaning that even if it's set to `False`, the profiler will still profiler symbolic operators. I have a issue about this: https://github.com/apache/incubator-mxnet/issues/15658. I think maybe we should have this flag removed in 2.0 to avoid confusion? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 commented on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0)
Zha0q1 commented on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0) URL: https://github.com/apache/incubator-mxnet/issues/9686#issuecomment-515231825 I would like to bring up one issue with profiler: currently, there is a flag `kSimbolic` that supposedly should control whether to profiler operators called in symbolic mode. However, there is rarely a use case where users would want to set it to `False`; also, in the backend, this flag is not used at all, meaning that even if it's set to `False`, the profiler will still profiler symbolic operators. I have a issue about this: https://github.com/apache/incubator-mxnet/issues/15658. I think maybe we should have this flag removed in 2.0? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larry77 commented on issue #15649: Installation Fails on Debian Stable Machine
larry77 commented on issue #15649: Installation Fails on Debian Stable Machine URL: https://github.com/apache/incubator-mxnet/issues/15649#issuecomment-515231136 Hi, And thanks! The installation from source eventually fails for the same reasons, look at this g++ -std=gnu++11 -shared -L/usr/lib/R/lib -Wl,-z,relro -o mxnet.so executor.o export.o im2rec.o io.o kvstore.o mxnet.o ndarray.o symbol.o -llapack -lblas -L/usr/lib/R/lib -lR make[1]: Leaving directory '/home/lorenzo/incubator-mxnet/R-package/src' installing to /usr/local/lib/R/site-library/00LOCK-R-package/00new/mxnet/libs ** R ** demo ** inst ** byte-compile and prepare package for lazy loading Note: ... may be used in an incorrect context ** help No man pages found in package ‘mxnet’ *** installing help indices ** building package indices ** installing vignettes ** testing if installed package can be loaded from temporary location [1] "Loading local: inst/libs/libmxnet.so" Error: package or namespace load failed for ‘mxnet’: .onLoad failed in loadNamespace() for 'mxnet', details: call: dyn.load("R-package/inst/libs/libmxnet.so", local = FALSE) error: unable to load shared object '/home/lorenzo/incubator-mxnet/R-package/R-package/inst/libs/libmxnet.so': /home/lorenzo/incubator-mxnet/R-package/R-package/inst/libs/libmxnet.so: cannot open shared object file: No such file or directory Error: loading failed Execution halted ERROR: loading failed * removing ‘/usr/local/lib/R/site-library/mxnet’ make: *** [Makefile:701: rpkg] Error 1 Banging my head against the wall This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 closed issue #15658: profile_symbolic flag not working in profiler
Zha0q1 closed issue #15658: profile_symbolic flag not working in profiler URL: https://github.com/apache/incubator-mxnet/issues/15658 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #15649: Installation Fails on Debian Stable Machine
zachgk commented on issue #15649: Installation Fails on Debian Stable Machine URL: https://github.com/apache/incubator-mxnet/issues/15649#issuecomment-515223474 Try running this which might fix the cuda.h problem. So, you are updating the default configuration instead of using the two lines instead of it (it doesn't load the default configuration if the config.mk file exists). ``` cp make/config.mk config.mk echo "USE_OPENCV = 1" >> ./config.mk echo "USE_BLAS = openblas" >> ./config.mk ``` @aaronmarkham or @anirudhacharya, do you know what is wrong with the R installation? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk opened a new pull request #15660: Bump Scala version to 1.6
zachgk opened a new pull request #15660: Bump Scala version to 1.6 URL: https://github.com/apache/incubator-mxnet/pull/15660 Fix API deprecation versions that were changed in #15072 @roywei @lanking520 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 edited a comment on issue #15069: Profiler RFC: Introducing new APIs
Zha0q1 edited a comment on issue #15069: Profiler RFC: Introducing new APIs URL: https://github.com/apache/incubator-mxnet/issues/15069#issuecomment-515213970 > optionally adding json output seems reasonable. during development of this feature, the intel folks were asking for more aggregation of the output, such as convolution ops for different input shapes, for instance, which seems perfectly fair (there’s a discussion issue about it somewhere). Anyway, whatever y’all decide to do, try to keep open the extensibility to stuff like that (which would lend itself more to json output than text output). However, I still think it’s important to have simple text output available if desired, since it gives a nice, clean one-look overview. Thanks for the tips! We have added support for sorting and json in this PR: https://github.com/apache/incubator-mxnet/pull/15132. Basically, we have added three new parameters to `dumps`, namely `sort_by`, `ascending`, and `format` by which you can control the aggregate stats output. We are keeping the old table view as it is clear to human; json format is intended for use cases where you want to parse the aggregate stats with a script This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 commented on issue #15069: Profiler RFC: Introducing new APIs
Zha0q1 commented on issue #15069: Profiler RFC: Introducing new APIs URL: https://github.com/apache/incubator-mxnet/issues/15069#issuecomment-515213970 > optionally adding json output seems reasonable. during development of this feature, the intel folks were asking for more aggregation of the output, such as convolution ops for different input shapes, for instance, which seems perfectly fair (there’s a discussion issue about it somewhere). Anyway, whatever y’all decide to do, try to keep open the extensibility to stuff like that (which would lend itself more to json output than text output). However, I still think it’s important to have simple text output available if desired, since it gives a nice, clean one-look overview. Thanks for the tips! We have added support for sorting and json in this PR: https://github.com/apache/incubator-mxnet/pull/15132. Basically, we have added three new parameters to `dumps`, namely `sort_by`, `ascending`, and `format` by which you can control the aggregate stats output. We are keeping the old table view as it is clear to human; json format is intended for use cases where you want to parse the aggregate stats atomatically This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] cjolivier01 commented on issue #15069: Profiler RFC: Introducing new APIs
cjolivier01 commented on issue #15069: Profiler RFC: Introducing new APIs URL: https://github.com/apache/incubator-mxnet/issues/15069#issuecomment-515211810 output was loosely based on pytorch, btw. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] larry77 commented on issue #15649: Installation Fails on Debian Stable Machine
larry77 commented on issue #15649: Installation Fails on Debian Stable Machine URL: https://github.com/apache/incubator-mxnet/issues/15649#issuecomment-515211654 Hello, And thanks for helping out. I have the files ~$ ls mxnet/lib/libmxnet.so -rwxr-xr-x 1 lorenzo lorenzo 71740832 Jul 24 17:25 mxnet/lib/libmxnet.so ~$ ls mxnet/R-package/inst/libs/libmxnet.so -rwxr-xr-x 1 root root 71740832 Jul 24 21:57 mxnet/R-package/inst/libs/libmxnet.so What I do not have is /home/lorenzo/mxnet/R-package/R-package/inst/libs/libmxnet.so because I do not have an R-package directory nested inside another R-package directory. This is the content of my R-package ls R-package/ total 40 drwxr-xr-x 2 lorenzo lorenzo 213 Jul 24 17:11 demo -rw-r--r-- 1 lorenzo lorenzo 1036 Jul 24 17:11 DESCRIPTION -rw-r--r-- 1 lorenzo lorenzo 507 Jul 24 17:11 dummy.NAMESPACE drwxr-xr-x 4 rootroot 45 Jul 24 18:06 inst -rw-r--r-- 1 lorenzo lorenzo 11350 Jul 24 17:11 LICENSE -rw-r--r-- 1 rootroot 520 Jul 25 22:45 NAMESPACE drwxr-xr-x 2 lorenzo lorenzo 4096 Jul 24 17:11 R -rw-r--r-- 1 lorenzo lorenzo 1170 Jul 24 17:11 README.md drwxr-xr-x 2 lorenzo lorenzo 4096 Jul 24 18:07 src drwxr-xr-x 3 lorenzo lorenzo30 Jul 24 17:11 tests drwxr-xr-x 2 lorenzo lorenzo 4096 Jul 24 17:11 vignettes And I did not alter anything by hand in the structure of the directory. All I did was git clone --recursive https://github.com/apache/incubator-mxnet.git mxnet cd mxnet/docs/install ./install_mxnet_ubuntu_python.sh ./install_mxnet_ubuntu_r.sh Could it be a trivial problem, meaning that everything is fine but for some reason R gets confused and looks for the library in the wrong place? ### As to building from source according to $ git clone --recursive https://github.com/apache/incubator-mxnet $ cd incubator-mxnet $ echo "USE_OPENCV = 1" >> ./config.mk $ echo "USE_BLAS = openblas" >> ./config.mk $ make -j $(nproc) I have a problem: the process fails due to the absence of cuda.h, but I would like to build for the CPU only. Any suggestion is very welcome! This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] cjolivier01 commented on issue #15069: Profiler RFC: Introducing new APIs
cjolivier01 commented on issue #15069: Profiler RFC: Introducing new APIs URL: https://github.com/apache/incubator-mxnet/issues/15069#issuecomment-515211361 optionally adding json output seems reasonable. during development of this feature, the intel folks were asking for more aggregation of the output, such as convolution ops for different input shapes, for instance, which seems perfectly fair (there’s a discussion issue about it somewhere). Anyway, whatever y’all decide to do, try to keep open the extensibility to stuff like that (which would lend itself more to json output than text output). However, I still think it’s important to have simple text output available if desired, since it gives a nice, clean one-look overview. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. marcoabreu pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 0e66503 Bump the publish timestamp. 0e66503 is described below commit 0e6650348b9ebe0545900393c6b6089cb654b7df Author: mxnet-ci AuthorDate: Thu Jul 25 20:55:12 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..2959c4e --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Thu Jul 25 20:55:12 UTC 2019
[GitHub] [incubator-mxnet] abhinavs95 commented on issue #15657: Eliminate common expressions
abhinavs95 commented on issue #15657: Eliminate common expressions URL: https://github.com/apache/incubator-mxnet/pull/15657#issuecomment-515208655 @mxnet-label-bot add [Backend, pr-awaiting-review] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. marcoabreu pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 5e86626c Bump the publish timestamp. 5e86626c is described below commit 5e86626cf5ad09ff0e9a9f9b3c3f62558c347612 Author: mxnet-ci AuthorDate: Thu Jul 25 19:16:10 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..415b778 --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Thu Jul 25 19:16:10 UTC 2019
[GitHub] [incubator-mxnet] zachgk commented on issue #15652: Using openblas
zachgk commented on issue #15652: Using openblas URL: https://github.com/apache/incubator-mxnet/issues/15652#issuecomment-515177387 I might be wrong. @larroy Can you take a look? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #15655: Performance regression for gluon dataloader with large batch size
zachgk commented on issue #15655: Performance regression for gluon dataloader with large batch size URL: https://github.com/apache/incubator-mxnet/issues/15655#issuecomment-515173584 Any idea @zhreshold? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 edited a comment on issue #15658: profile_symbolic flag not working in profiler
Zha0q1 edited a comment on issue #15658: profile_symbolic flag not working in profiler URL: https://github.com/apache/incubator-mxnet/issues/15658#issuecomment-515165813 I looked into the code. It seemed like even though `profile_symbolic` was defined, it was never used anywhere. @anirudh2290 do you know about this? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 commented on issue #15658: profile_symbolic flag not working in profiler
Zha0q1 commented on issue #15658: profile_symbolic flag not working in profiler URL: https://github.com/apache/incubator-mxnet/issues/15658#issuecomment-515165813 I looked into the code. I seemed like even though `profile_symbolic` was defined, it was never used anywhere. @anirudh2290 do you know about this? This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] gogazago commented on issue #15652: Using openblas
gogazago commented on issue #15652: Using openblas URL: https://github.com/apache/incubator-mxnet/issues/15652#issuecomment-515163120 so can we ignore openblas on https://github.com/apache/incubator-mxnet/blob/master/Makefile#L217 ? Thanks, Sean. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zachgk commented on issue #15652: Using openblas
zachgk commented on issue #15652: Using openblas URL: https://github.com/apache/incubator-mxnet/issues/15652#issuecomment-515161904 We have it as a switch (see https://github.com/apache/incubator-mxnet/blob/master/Makefile#L255). You should have USE_BLAS=open to use openblas This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] IvyBazan commented on issue #15580: update profiler tutorial
IvyBazan commented on issue #15580: update profiler tutorial URL: https://github.com/apache/incubator-mxnet/pull/15580#issuecomment-515161350 LGTM! This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] matteosal opened a new issue #15659: Dropout produces wrong mask with MKL-DNN
matteosal opened a new issue #15659: Dropout produces wrong mask with MKL-DNN URL: https://github.com/apache/incubator-mxnet/issues/15659 ## Description Dropout mask (second output) is always full of zeros when mxnet is built with MKL-DNN. Weirdly, setting `MXNET_MKLDNN_ENABLED=0` doesn't fix it, but using a build without MKL-DNN does. ## Environment info (Required) ``` --Python Info-- Version : 3.7.2 Compiler : GCC 7.3.0 Build: ('default', 'Dec 29 2018 06:19:36') Arch : ('64bit', '') Pip Info--- Version : 19.0.1 Directory: /opt/Anaconda/lib/python3.7/site-packages/pip --MXNet Info--- Version : 1.5.0 Directory: /home/matteo/Git/mxnet/python/mxnet Commit hash file "/home/matteo/Git/mxnet/python/mxnet/COMMIT_HASH" not found. Not installed from pre-built package or built from source. Library : ['/home/matteo/Git/mxnet/python/mxnet/../../lib/libmxnet.so'] Build features: ✖ CUDA ✖ CUDNN ✖ NCCL ✖ CUDA_RTC ✖ TENSORRT ✔ CPU_SSE ✔ CPU_SSE2 ✔ CPU_SSE3 ✔ CPU_SSE4_1 ✔ CPU_SSE4_2 ✖ CPU_SSE4A ✔ CPU_AVX ✖ CPU_AVX2 ✖ OPENMP ✖ SSE ✔ F16C ✔ JEMALLOC ✖ BLAS_OPEN ✔ BLAS_ATLAS ✖ BLAS_MKL ✖ BLAS_APPLE ✖ LAPACK ✔ MKLDNN ✖ OPENCV ✖ CAFFE ✖ PROFILER ✖ DIST_KVSTORE ✖ CXX14 ✖ INT64_TENSOR_SIZE ✖ SIGNAL_HANDLER ✖ DEBUG --System Info-- Platform : Linux-4.15.0-55-generic-x86_64-with-debian-buster-sid system : Linux node : mongolius release : 4.15.0-55-generic version : #60-Ubuntu SMP Tue Jul 2 18:22:20 UTC 2019 --Hardware Info-- machine : x86_64 processor: x86_64 Architecture:x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 8 On-line CPU(s) list: 0-7 Thread(s) per core: 2 Core(s) per socket: 4 Socket(s): 1 NUMA node(s):1 Vendor ID: GenuineIntel CPU family: 6 Model: 94 Model name: Intel(R) Core(TM) i7-6700HQ CPU @ 2.60GHz Stepping:3 CPU MHz: 2700.253 CPU max MHz: 3500, CPU min MHz: 800, BogoMIPS:5184.00 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache:256K L3 cache:6144K NUMA node0 CPU(s): 0-7 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single pti ssbd ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm mpx rdseed adx smap clflushopt intel_pt xsaveopt xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp md_clear flush_l1d --Network Test-- Setting timeout: 10 Timing for MXNet: https://github.com/apache/incubator-mxnet, DNS: 0.0117 sec, LOAD: 0.8935 sec. Timing for Gluon Tutorial(en): http://gluon.mxnet.io, DNS: 0.0599 sec, LOAD: 2.1901 sec. Timing for Gluon Tutorial(cn): https://zh.gluon.ai, DNS: 0.1028 sec, LOAD: 0.9832 sec. Timing for FashionMNIST: https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/dataset/fashion-mnist/train-labels-idx1-ubyte.gz, DNS: 0.0657 sec, LOAD: 1.2597 sec. Timing for PYPI: https://pypi.python.org/pypi/pip, DNS: 0.0380 sec, LOAD: 0.8543 sec. Timing for Conda: https://repo.continuum.io/pkgs/free/, DNS: 0.0395 sec, LOAD: 0.4625 sec. ``` Using python package. ## Build info (Required if built from source) Compiler: gcc MXNet commit hash: 076b2f330c60f05cb939beea28dd04cd571a34c0 Build config: plain `config.mk`, except for `USE_OPENCV=0` ## Minimum reproducible example ``` import mxnet as mx json = '{\"nodes\":[{\"op\":\"null\",\"name\":\"in\",\"inputs\":[]},{\"op\":\ \"Dropout\",\"name\":\"dropout\",\"attrs\":{\"axes\":\"()\",\"mode\":\ \"always\",\"p\":\"0.2\"},\"inputs\":[[0,0,0]]}],\"arg_nodes\":[0],\"\ heads\":[[1,0,0],[1,1,0]]}' sym = mx.sym.load_json(json) ex = sym.bind(mx.cpu(), {'in': mx.nd.array(range(10))}) ex.forward() print(ex.outputs) ``` When running the above script using a build with MKL-DNN, the second output is full of `1.e-45`. When using a build without MKL-DNN, it is as expected. A side question: if I were to build that symbol using
[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15659: Dropout produces wrong mask with MKL-DNN
mxnet-label-bot commented on issue #15659: Dropout produces wrong mask with MKL-DNN URL: https://github.com/apache/incubator-mxnet/issues/15659#issuecomment-515159883 Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended labels: Build This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya edited a comment on issue #14617: PDF operators for the random samplers, and also the Dirichlet
ChaiBapchya edited a comment on issue #14617: PDF operators for the random samplers, and also the Dirichlet URL: https://github.com/apache/incubator-mxnet/pull/14617#issuecomment-515157689 ~~Curious to know if pdf operators aren't supposed to have NDArray API? Currently they're only in Symbol API right? @sxjscience @david-seiler~~ Realized it has an NDArray API. Just not mentioned in the docs. Thanks. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ChaiBapchya commented on issue #14617: PDF operators for the random samplers, and also the Dirichlet
ChaiBapchya commented on issue #14617: PDF operators for the random samplers, and also the Dirichlet URL: https://github.com/apache/incubator-mxnet/pull/14617#issuecomment-515157689 Curious to know if pdf operators aren't supposed to have NDArray API? Currently they're only in Symbol API right? @sxjscience @david-seiler Thanks. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 commented on issue #15658: profile_symbolic flag not working in profiler
Zha0q1 commented on issue #15658: profile_symbolic flag not working in profiler URL: https://github.com/apache/incubator-mxnet/issues/15658#issuecomment-515153365 @mxnet-label-bot add [Profiler, Bug] This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ptrendx opened a new pull request #15657: Eliminate common expressions
ptrendx opened a new pull request #15657: Eliminate common expressions URL: https://github.com/apache/incubator-mxnet/pull/15657 ## Description ## This PR introduces a graph pass that eliminates redundant common expressions in the graph. For example, let's look at the graph created from a following Python code: ``` a = mx.sym.Variable() out = (a + 5) * (a + 5) ``` This graph will compute `a+5` twice, which is wasteful. After the pass introduced in this PR, the actual executed graph will be equivalent to ``` a = mx.sym.Variable() b = a + 5 out = b * b ``` which computes the `a+5` only once. @eric-haibin-lin FYI ## Checklist ## ### Essentials ### Please feel free to remove inapplicable items for your PR. - [x] Changes are complete (i.e. I finished coding on this PR) - [ ] All changes have test coverage: - Unit tests are added for small changes to verify correctness (e.g. adding a new operator) - [x] Code is well-documented: - For new C++ functions in header files, their functionalities and arguments are documented. - [x] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ### Changes ### - [ ] Feature1, tests, (and when applicable, API doc) - [ ] Feature2, tests, (and when applicable, API doc) ## Comments ## This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Zha0q1 opened a new issue #15658: profile_symbolic flag not working in profiler
Zha0q1 opened a new issue #15658: profile_symbolic flag not working in profiler URL: https://github.com/apache/incubator-mxnet/issues/15658 We use `profiler.set_config()` to control the profiler's behaviors, i.e. what to profile. `profile_symbolic` and `profile_imperative` are two parameters to control whether to profile operators called in the two modes respectively. However, it seems currently `profiler_symbolic` is not functioning correctly. In the screenshot below, we have a simple scrip that has `plus_scalar` in symbolic mode and `minus_scalar` in imperative mode. `profile_symbolic` and `profile_imperative` are both set to `True` and we do see both events in `profiler.dumps()`. ![Screen Shot 2019-07-25 at 10 55 47 AM](https://user-images.githubusercontent.com/16669457/61897220-78376080-aecb-11e9-82dd-68cff9b83c6d.png) However, if we set `profile_symbolic` and `profile_imperative` to `False`. The `minus_sclar` event is gone, but we still have the `plus_scalar` event, which is not the expected behavior. ![Screen Shot 2019-07-25 at 10 56 16 AM](https://user-images.githubusercontent.com/16669457/61897214-753c7000-aecb-11e9-89a6-e876b0b6ccb3.png) MXNet version: from 1.4 to current (I only tested as far as 1.4) Python version: 2&3 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15658: profile_symbolic flag not working in profiler
mxnet-label-bot commented on issue #15658: profile_symbolic flag not working in profiler URL: https://github.com/apache/incubator-mxnet/issues/15658#issuecomment-515152798 Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended labels: Bug This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] sandeep-krishnamurthy closed issue #14100: More user-friendly profiler result
sandeep-krishnamurthy closed issue #14100: More user-friendly profiler result URL: https://github.com/apache/incubator-mxnet/issues/14100 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] sandeep-krishnamurthy closed issue #15069: Profiler RFC: Introducing new APIs
sandeep-krishnamurthy closed issue #15069: Profiler RFC: Introducing new APIs URL: https://github.com/apache/incubator-mxnet/issues/15069 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] sandeep-krishnamurthy closed issue #15557: mxnet.profiler.dump creates invalid json files
sandeep-krishnamurthy closed issue #15557: mxnet.profiler.dump creates invalid json files URL: https://github.com/apache/incubator-mxnet/issues/15557 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] sandeep-krishnamurthy closed issue #10504: Can MXNet operator profiling work well using gluon model?
sandeep-krishnamurthy closed issue #10504: Can MXNet operator profiling work well using gluon model? URL: https://github.com/apache/incubator-mxnet/issues/10504 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] sandeep-krishnamurthy closed issue #15556: mxnet profiler.dump error: MXNetError(py_str(_LIB.MXGetLastError()))
sandeep-krishnamurthy closed issue #15556: mxnet profiler.dump error: MXNetError(py_str(_LIB.MXGetLastError())) URL: https://github.com/apache/incubator-mxnet/issues/15556 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] sandeep-krishnamurthy closed issue #15486: mxnet profiler yields exception when multithreaded mxndarray IO operations occur in the background
sandeep-krishnamurthy closed issue #15486: mxnet profiler yields exception when multithreaded mxndarray IO operations occur in the background URL: https://github.com/apache/incubator-mxnet/issues/15486 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] marcoabreu commented on issue #15637: Fix _copy_to on MKLDNN backend
marcoabreu commented on issue #15637: Fix _copy_to on MKLDNN backend URL: https://github.com/apache/incubator-mxnet/pull/15637#issuecomment-515133906 Ok, feel free to go ahead. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307404045 ## File path: example/lib_api/libtest.cc ## @@ -0,0 +1,77 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +/*! + * Copyright (c) 2015 by Contributors + * \file libtest.cc + * \brief Tests if the library is implemented correctly + */ + +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) +#include +#else +#include +#endif + +#include +#include "../../include/mxnet/lib_api.h" + +#define MXNET_VERSION 10500 + +int main(void) { + // Get a handle to the library. +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) + HINSTANCE handle; + handle = LoadLibrary(TEXT("lib.dll")); Review comment: this is called mylib.dll This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support
samskalicky commented on a change in pull request #15489: [WIP] Dynamic Library Loading Support URL: https://github.com/apache/incubator-mxnet/pull/15489#discussion_r307402120 ## File path: example/lib_api/libtest.cc ## @@ -0,0 +1,77 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +/*! + * Copyright (c) 2015 by Contributors + * \file libtest.cc + * \brief Tests if the library is implemented correctly + */ + +#if defined(_WIN32) || defined(_WIN64) || defined(__WINDOWS__) +#include +#else +#include +#endif + +#include +#include "../../include/mxnet/lib_api.h" Review comment: lets just change this to "lib_api.h" so it matches the mylib.cc file below. You already have the include path setup in the makefile as: -I ../../include/mxnet so we dont need absolute path in the include statement This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] kshitij12345 edited a comment on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
kshitij12345 edited a comment on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656#issuecomment-515095374 The documentation seems to be incomplete. https://mxnet.incubator.apache.org/api/python/ndarray/ndarray.html#mxnet.ndarray.UpSampling It doesn't mention about weight when using `bilinear`. Digging into the code, https://github.com/apache/incubator-mxnet/blob/8158ba4b0f1ebd696ec09a0b1aa6031bacb60740/src/operator/nn/upsampling.cc#L123-L173 It mentions about requiring the weight and also gives an example. I have verified that the example works, ``` from mxnet import nd x = nd.ones((1,3,3,3)) # NCHW w = nd.ones((3,1,4,4)) upsampled = nd.UpSampling(x, w, scale=2, sample_type='bilinear', num_filter=3) ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] kshitij12345 commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
kshitij12345 commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656#issuecomment-515095374 There seems to be a problem with the documentation. https://mxnet.incubator.apache.org/api/python/ndarray/ndarray.html#mxnet.ndarray.UpSampling It doesn't mention about weight when using `bilinear`. Digging into the code, https://github.com/apache/incubator-mxnet/blob/8158ba4b0f1ebd696ec09a0b1aa6031bacb60740/src/operator/nn/upsampling.cc#L123-L173 It mentions about requiring the weight and also gives an example. I have verified that the example works, ``` from mxnet import nd x = nd.ones((1,3,3,3)) # NCHW w = nd.ones((3,1,4,4)) upsampled = nd.UpSampling(x, w, scale=2, sample_type='bilinear', num_filter=3) ``` This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
mxnet-label-bot commented on issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656#issuecomment-515048903 Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended labels: Test This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] ngunauj opened a new issue #15656: Can you give me a UpSampling's doc? 'bilinear'!!
ngunauj opened a new issue #15656: Can you give me a UpSampling's doc? 'bilinear'!! URL: https://github.com/apache/incubator-mxnet/issues/15656 Note: Providing complete information in the most concise form is the best way to get help. This issue template serves as the checklist for essential information to most of the technical issues and bug reports. For non-technical issues and feature requests, feel free to present the information in what you believe is the best form. For Q & A and discussion, please start a discussion thread at https://discuss.mxnet.io ## Description (Brief description of the problem in no more than 2 sentences.) ## Environment info (Required) ``` What to do: 1. Download the diagnosis script from https://raw.githubusercontent.com/apache/incubator-mxnet/master/tools/diagnose.py 2. Run the script using `python diagnose.py` and paste its output here. ``` Package used (Python/R/Scala/Julia): (I'm using ...) For Scala user, please provide: 1. Java version: (`java -version`) 2. Maven version: (`mvn -version`) 3. Scala runtime if applicable: (`scala -version`) For R user, please provide R `sessionInfo()`: ## Build info (Required if built from source) Compiler (gcc/clang/mingw/visual studio): MXNet commit hash: (Paste the output of `git rev-parse HEAD` here.) Build config: (Paste the content of config.mk, or the build command.) ## Error Message: (Paste the complete error message, including stack trace.) ## Minimum reproducible example (If you are using your own code, please provide a short script that reproduces the error. Otherwise, please provide link to the existing example.) ## Steps to reproduce (Paste the commands you ran that produced the error.) 1. 2. ## What have you tried to solve it? 1. 2. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. marcoabreu pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 58be935 Bump the publish timestamp. 58be935 is described below commit 58be9359c5bf6adac57fbb37697311233927042d Author: mxnet-ci AuthorDate: Thu Jul 25 13:17:24 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..3d574ed --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Thu Jul 25 13:17:24 UTC 2019
[GitHub] [incubator-mxnet] Neutron3529 edited a comment on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0)
Neutron3529 edited a comment on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0) URL: https://github.com/apache/incubator-mxnet/issues/9686#issuecomment-514969292 (1)#10840 Add einsum since it is useful and it could simplify linear algebra ops. (it is now supported by tensorflow and pyThrch) (2)seperate MXNet.dll into small files to reduce the import time. nvcc will generate a HUGE file with `-arch=(...a lot of sm_*...)` it seems seperate the HUGE `KNOWN_CUDA_ARCHS` instruction may increase the `import` performance. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] Neutron3529 commented on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0)
Neutron3529 commented on issue #9686: [Discussion] MXNet 2.0 Roadmap (was: APIs that might be a good idea to break in 2.0) URL: https://github.com/apache/incubator-mxnet/issues/9686#issuecomment-514969292 (1)#10840 Add einsum since it is useful and it could simplify linear algebra ops. (it is now supported by tensorflow and pyThrch) (2)seperate MXNet.dll into small files to reduce the import time. nvcc will generate a HUGE file with `-sm=(...a lot of codes...)` it seems seperate the HUGE `-sm` instruction may increase the `import` performance. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] turtlemeow commented on issue #13434: Inference with C++ using gpu on tx2 got error
turtlemeow commented on issue #13434: Inference with C++ using gpu on tx2 got error URL: https://github.com/apache/incubator-mxnet/issues/13434#issuecomment-514960538 @chenjunweii thanks! Input a smaller image and it works, thank you This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] chenjunweii commented on issue #13434: Inference with C++ using gpu on tx2 got error
chenjunweii commented on issue #13434: Inference with C++ using gpu on tx2 got error URL: https://github.com/apache/incubator-mxnet/issues/13434#issuecomment-514927278 @turtlemeow I think the problem might be out of the memory, try smaller input size, memory management of python api and c++ api seems different This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-mxnet-site] branch asf-site updated: Bump the publish timestamp.
This is an automated email from the ASF dual-hosted git repository. marcoabreu pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/incubator-mxnet-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 18486d3 Bump the publish timestamp. 18486d3 is described below commit 18486d3b07324794292f223299cf9fffbbd702a7 Author: mxnet-ci AuthorDate: Thu Jul 25 07:16:08 2019 + Bump the publish timestamp. --- date.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/date.txt b/date.txt new file mode 100644 index 000..1ba87e7 --- /dev/null +++ b/date.txt @@ -0,0 +1 @@ +Thu Jul 25 07:16:08 UTC 2019
[GitHub] [incubator-mxnet] mxnet-label-bot commented on issue #15655: Performance regression for gluon dataloader with large batch size
mxnet-label-bot commented on issue #15655: Performance regression for gluon dataloader with large batch size URL: https://github.com/apache/incubator-mxnet/issues/15655#issuecomment-514914519 Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended labels: Gluon, Performance This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services