[GitHub] pengzhao-intel commented on issue #10075: Fix CMake build issue with MKL.

pengzhao-intel commented on issue #10075: Fix CMake build issue with MKL. URL: https://github.com/apache/incubator-mxnet/pull/10075#issuecomment-372565487 > Not sure. At this point, I really don't know what's going on with the mkl build, what works with what, does mkldnn use mklml or mkl? c

[GitHub] sxjscience commented on issue #10085: Ordering operators do not support kNullOp

sxjscience commented on issue #10085: Ordering operators do not support kNullOp URL: https://github.com/apache/incubator-mxnet/issues/10085#issuecomment-372565456 @chinakook You can get both values using the following code: ```python values, inds = mx.sym.topk(a, axis=-1, k=5, ret_typ

[GitHub] chinakook commented on issue #10085: Ordering operators do not support kNullOp

chinakook commented on issue #10085: Ordering operators do not support kNullOp URL: https://github.com/apache/incubator-mxnet/issues/10085#issuecomment-372564930 Yes, It's only support indices only. Thanks for your help! This

[GitHub] sxjscience commented on issue #10085: Ordering operators do not support kNullOp

sxjscience commented on issue #10085: Ordering operators do not support kNullOp URL: https://github.com/apache/incubator-mxnet/issues/10085#issuecomment-372563935 @chinakook I'll fix it tomorrow. This is an automated message

[GitHub] sxjscience commented on issue #10085: Ordering operators do not support kNullOp

sxjscience commented on issue #10085: Ordering operators do not support kNullOp URL: https://github.com/apache/incubator-mxnet/issues/10085#issuecomment-372563598 The error also occurs in other ordering operators like Sort, ArgSort. https://github.com/apache/incubator-mxnet/blob/master/src

[GitHub] pengzhao-intel commented on issue #8974: MXNET compatibility with MKL libraries bundled in Microsoft R Open

pengzhao-intel commented on issue #8974: MXNET compatibility with MKL libraries bundled in Microsoft R Open URL: https://github.com/apache/incubator-mxnet/issues/8974#issuecomment-372562963 @mjmg The MKLML has been replaced by MKL-DNN. Could you try again?

[GitHub] pengzhao-intel commented on issue #8881: Intel MKL FATAL ERROR: Cannot load libmkl_avx512_mic.so or libmkl_def.so.

pengzhao-intel commented on issue #8881: Intel MKL FATAL ERROR: Cannot load libmkl_avx512_mic.so or libmkl_def.so. URL: https://github.com/apache/incubator-mxnet/issues/8881#issuecomment-372562280 Resolved by #9112 and the mklml has been replaced by MKL-DNN. This should be closed. @sand

[GitHub] pengzhao-intel commented on issue #8881: Intel MKL FATAL ERROR: Cannot load libmkl_avx512_mic.so or libmkl_def.so.

pengzhao-intel commented on issue #8881: Intel MKL FATAL ERROR: Cannot load libmkl_avx512_mic.so or libmkl_def.so. URL: https://github.com/apache/incubator-mxnet/issues/8881#issuecomment-372562280 Resolved by #9112 and the mklml has been replaced by MKL-DNN. This should be closed. @sand

[GitHub] sxjscience commented on issue #10085: Topk op cannot run!

sxjscience commented on issue #10085: Topk op cannot run! URL: https://github.com/apache/incubator-mxnet/issues/10085#issuecomment-372562351 It's because the first argument is set as `_`, which gives `req[0]=kNullOp`. If you only needs the indices, a temporary workaround is to use `inds =

[GitHub] pengzhao-intel commented on issue #8874: mxnet installation from source: C++ linkage error on HPC

pengzhao-intel commented on issue #8874: mxnet installation from source: C++ linkage error on HPC URL: https://github.com/apache/incubator-mxnet/issues/8874#issuecomment-372561464 @jerrin92 does the issue resolved? This is a

[GitHub] chinakook opened a new issue #10085: Topk op cannot run!

chinakook opened a new issue #10085: Topk op cannot run! URL: https://github.com/apache/incubator-mxnet/issues/10085 When testing ```python a = mx.sym.var("data", shape=(6,7,8)) adata= nd.random.uniform(shape=(6,7,8), ctx=mx.gpu(0)) _, inds = mx.sym.topk(a, axis=-1, k=5, ret_typ=

[GitHub] eric-haibin-lin commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd?

eric-haibin-lin commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd? URL: https://github.com/apache/incubator-mxnet/issues/10084#issuecomment-372560412 Also sparse adagrad. Note that GPU optimizer performance can be further improved if applied with s

[GitHub] moveforever commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd?

moveforever commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd? URL: https://github.com/apache/incubator-mxnet/issues/10084#issuecomment-372557589 Just a question on what sparse optimizer was implemented, tks

[GitHub] eric-haibin-lin commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd?

eric-haibin-lin commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd? URL: https://github.com/apache/incubator-mxnet/issues/10084#issuecomment-372556099 Is it a question on what sparse optimizer was implemented, or what welcomes contribution? --

[GitHub] pengzhao-intel commented on issue #8532: mxnet-mkl (v0.12.0) crash when using (conda-installed) numpy with MKL

pengzhao-intel commented on issue #8532: mxnet-mkl (v0.12.0) crash when using (conda-installed) numpy with MKL URL: https://github.com/apache/incubator-mxnet/issues/8532#issuecomment-372555255 @fhieber could you try the new build of 1.1.0 https://github.com/apache/incubator-mxnet/releas

[GitHub] pengzhao-intel commented on issue #8532: mxnet-mkl (v0.12.0) crash when using (conda-installed) numpy with MKL

pengzhao-intel commented on issue #8532: mxnet-mkl (v0.12.0) crash when using (conda-installed) numpy with MKL URL: https://github.com/apache/incubator-mxnet/issues/8532#issuecomment-372555255 @fhieber could you try the new build of 1.1.0 https://github.com/apache/incubator-mxnet/releas

[GitHub] sxjscience commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd?

sxjscience commented on issue #10084: which optimizer to support sparse update except adam, ftrl, sgd? URL: https://github.com/apache/incubator-mxnet/issues/10084#issuecomment-372554612 @eric-haibin-lin @ZiyueHuang This is

[GitHub] moveforever opened a new issue #10084: which optimizer to support sparse update except adam, ftrl, sgd?

moveforever opened a new issue #10084: which optimizer to support sparse update except adam, ftrl, sgd? URL: https://github.com/apache/incubator-mxnet/issues/10084 Note: Providing complete information in the most concise form is the best way to get help. This issue template serves as the c

[GitHub] sxjscience commented on issue #10035: Who can help me solve this error??batch_loss.backward() error?

sxjscience commented on issue #10035: Who can help me solve this error??batch_loss.backward() error? URL: https://github.com/apache/incubator-mxnet/issues/10035#issuecomment-372553460 @kenjewu May I know the status of this question? ---

[GitHub] sxjscience commented on a change in pull request #10000: [MXNET-80] Fix average pooling kernel size assignment error

sxjscience commented on a change in pull request #1: [MXNET-80] Fix average pooling kernel size assignment error URL: https://github.com/apache/incubator-mxnet/pull/1#discussion_r174021284 ## File path: src/operator/nn/pooling.cc ## @@ -54,11 +54,13 @@ static void

[GitHub] tqchen commented on issue #10083: [TENSOR] Fix DLTensor conversion for int64

tqchen commented on issue #10083: [TENSOR] Fix DLTensor conversion for int64 URL: https://github.com/apache/incubator-mxnet/pull/10083#issuecomment-372550931 cc @ZihengJiang This is an automated message from the Apache Git Se

[GitHub] sxjscience commented on a change in pull request #10000: [MXNET-80] Fix average pooling kernel size assignment error

sxjscience commented on a change in pull request #1: [MXNET-80] Fix average pooling kernel size assignment error URL: https://github.com/apache/incubator-mxnet/pull/1#discussion_r174019453 ## File path: src/operator/nn/pooling-inl.h ## @@ -56,11 +56,11 @@ struct Po

[GitHub] tqchen opened a new pull request #10083: [TENSOR] Fix DLTensor conversion for int64

tqchen opened a new pull request #10083: [TENSOR] Fix DLTensor conversion for int64 URL: https://github.com/apache/incubator-mxnet/pull/10083 This is a bugfix PR that fixes the DLTensor conversion for type int64, which is not covered by previous test. I have updated the test to cover this

[GitHub] CoinCheung commented on a change in pull request #10000: [MXNET-80] Fix average pooling kernel size assignment error

CoinCheung commented on a change in pull request #1: [MXNET-80] Fix average pooling kernel size assignment error URL: https://github.com/apache/incubator-mxnet/pull/1#discussion_r174018923 ## File path: src/operator/nn/pooling-inl.h ## @@ -56,11 +56,11 @@ struct Po

[GitHub] haojin2 commented on a change in pull request #10078: Support float16 in L2Normalization operator

haojin2 commented on a change in pull request #10078: Support float16 in L2Normalization operator URL: https://github.com/apache/incubator-mxnet/pull/10078#discussion_r174012945 ## File path: src/operator/l2_normalization.cc ## @@ -26,13 +26,22 @@ namespace mxnet { names

[GitHub] cjolivier01 commented on a change in pull request #10078: Support float16 in L2Normalization operator

cjolivier01 commented on a change in pull request #10078: Support float16 in L2Normalization operator URL: https://github.com/apache/incubator-mxnet/pull/10078#discussion_r174012691 ## File path: src/operator/l2_normalization.cc ## @@ -26,13 +26,22 @@ namespace mxnet { n

[GitHub] haojin2 commented on issue #10078: Support float16 in L2Normalization operator

haojin2 commented on issue #10078: Support float16 in L2Normalization operator URL: https://github.com/apache/incubator-mxnet/pull/10078#issuecomment-372540444 I think this PR should be ready for merge, @rahul003 would you please take a look at it to double-check? Thanks! -

[GitHub] cjolivier01 commented on issue #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on issue #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#issuecomment-372538789 how well you learned the profiler stuff so fast is impressive -

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010079 ## File path: include/mxnet/kvstore.h ## @@ -361,6 +373,1

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010410 ## File path: src/kvstore/kvstore_dist_server.h ## @@ -32

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010585 ## File path: src/kvstore/kvstore_dist_server.h ## @@ -17

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174009887 ## File path: example/image-classification/common/fit.py #

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010221 ## File path: include/mxnet/kvstore.h ## @@ -361,6 +373,1

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010394 ## File path: src/kvstore/kvstore_dist_server.h ## @@ -32

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010534 ## File path: src/kvstore/kvstore_dist_server.h ## @@ -12

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010819 ## File path: src/kvstore/kvstore_dist_server.h ## @@ -15

[GitHub] cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

cjolivier01 commented on a change in pull request #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#discussion_r174010353 ## File path: src/kvstore/kvstore_dist_server.h ## @@ -32

[GitHub] feevos commented on issue #2910: Helper function to get GPU device number?

feevos commented on issue #2910: Helper function to get GPU device number? URL: https://github.com/apache/incubator-mxnet/issues/2910#issuecomment-372536394 @kenfehling I get the same error when I run this command on my laptop (single GPU) but not on the HPC cluster. Somehow the key ```

[GitHub] szha commented on issue #9705: Added unittest for benchmarking metric performance

szha commented on issue #9705: Added unittest for benchmarking metric performance URL: https://github.com/apache/incubator-mxnet/pull/9705#issuecomment-372535302 One last request: would you put the performance tests in a separate test file, such as test_metric_perf.py, so that it's easier

[GitHub] rahul003 commented on issue #9933: [MXNET-23] Adding support to profile kvstore server during distributed training

rahul003 commented on issue #9933: [MXNET-23] Adding support to profile kvstore server during distributed training URL: https://github.com/apache/incubator-mxnet/pull/9933#issuecomment-372531644 @cjolivier01 (and others), what do you think? -

[GitHub] asitstands commented on a change in pull request #10048: [MXNET-68] Random shuffle implementation

asitstands commented on a change in pull request #10048: [MXNET-68] Random shuffle implementation URL: https://github.com/apache/incubator-mxnet/pull/10048#discussion_r173857649 ## File path: src/operator/random/shuffle_op.cc ## @@ -0,0 +1,134 @@ +/* + * Licensed to the Ap

[GitHub] asitstands commented on a change in pull request #10048: [MXNET-68] Random shuffle implementation

asitstands commented on a change in pull request #10048: [MXNET-68] Random shuffle implementation URL: https://github.com/apache/incubator-mxnet/pull/10048#discussion_r173857649 ## File path: src/operator/random/shuffle_op.cc ## @@ -0,0 +1,134 @@ +/* + * Licensed to the Ap

[GitHub] cjolivier01 commented on issue #10042: Gluon dataloader crash on speech recognition training

cjolivier01 commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372522379 ok, will take a look tomorrow On Mon, Mar 12, 2018 at 7:03 PM Xingjian Shi wrote: > 100% >

[GitHub] sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training

sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372521638 100% Get Outlook for iOS From: Chris Olivier

[GitHub] cjolivier01 commented on issue #10042: Gluon dataloader crash on speech recognition training

cjolivier01 commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372518547 with what frequency does it occur? This is an au

[GitHub] sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training

sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372517195 @cjolivier01 @piiswrong After BinarySearch, I can confirm that the problem is due to this PR: https://github.com/

[GitHub] sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training

sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372517195 @cjolivier01 @piiswrong After BinarySearch, I can confirm that the problem is due to this PR: https://github.com/

[GitHub] sxjscience commented on issue #10082: All workloads are pushed to the key queue in multi-processing DataLoader

sxjscience commented on issue #10082: All workloads are pushed to the key queue in multi-processing DataLoader URL: https://github.com/apache/incubator-mxnet/issues/10082#issuecomment-372514596 @piiswrong @yzhliu This is an

[GitHub] sxjscience opened a new issue #10082: All workloads are pushed to the key queue in multi-processing DataLoader

sxjscience opened a new issue #10082: All workloads are pushed to the key queue in multi-processing DataLoader URL: https://github.com/apache/incubator-mxnet/issues/10082 In the current implementation of the multi-processing part of the DataLoader, all batches in the `batch_sampler` are pu

[GitHub] CoinCheung commented on a change in pull request #10000: [MXNET-80] Fix average pooling kernel size assignment error

CoinCheung commented on a change in pull request #1: [MXNET-80] Fix average pooling kernel size assignment error URL: https://github.com/apache/incubator-mxnet/pull/1#discussion_r173990770 ## File path: src/operator/nn/pooling-inl.h ## @@ -56,11 +56,11 @@ struct Po

[GitHub] xinyu-intel commented on a change in pull request #9918: [MXNET-74]Update mkldnn to the newest & Add clang build test with mkldnn.

xinyu-intel commented on a change in pull request #9918: [MXNET-74]Update mkldnn to the newest & Add clang build test with mkldnn. URL: https://github.com/apache/incubator-mxnet/pull/9918#discussion_r173990615 ## File path: Jenkinsfile ## @@ -175,6 +175,24 @@ try {

[GitHub] CoinCheung commented on a change in pull request #10000: [MXNET-80] Fix average pooling kernel size assignment error

CoinCheung commented on a change in pull request #1: [MXNET-80] Fix average pooling kernel size assignment error URL: https://github.com/apache/incubator-mxnet/pull/1#discussion_r173990460 ## File path: src/operator/nn/pooling-inl.h ## @@ -56,11 +56,11 @@ struct Po

[GitHub] szha commented on a change in pull request #10025: Language model with Google's billion words dataset

szha commented on a change in pull request #10025: Language model with Google's billion words dataset URL: https://github.com/apache/incubator-mxnet/pull/10025#discussion_r173988730 ## File path: example/rnn/large_word_lm/data.py ## @@ -0,0 +1,202 @@ +# Licensed to the Apa

[GitHub] sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training

sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372504574 I'm now doing a binary search to locate the problem. -

[GitHub] ShootingSpace commented on issue #10068: rnn.encode_sentences deals with unknown token

ShootingSpace commented on issue #10068: rnn.encode_sentences deals with unknown token URL: https://github.com/apache/incubator-mxnet/issues/10068#issuecomment-372504273 This is a simple but necessary solution. As people sometimes are only interested in k most frequent tokens, that is the

[GitHub] ShootingSpace commented on issue #10068: rnn.encode_sentences deals with unknown token

ShootingSpace commented on issue #10068: rnn.encode_sentences deals with unknown token URL: https://github.com/apache/incubator-mxnet/issues/10068#issuecomment-372504273 This is a simple but necessary solution. As people sometimes are only interested in k most frequent token, that is the

[GitHub] sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training

sxjscience commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372501972 @Jerryzcn found that v1.1.0 does not have this problem. -

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372493368 Ok, I can reproduce with CUDNN enabled... [Epoch 0] accuracy=0.101000 [Epoch 0]

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372493368 Ok, I can reproduce with CUDNN enabled... ---

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372490007 I didn't have CUDNN enabled. Trying that now... -

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372489626 Also, is this done with the latest build? ---

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372489626 Also, is this doen with the latest build? ---

[GitHub] fedorzh commented on issue #9410: Training with the same parameters and seed gets significantly different results

fedorzh commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372489419 Amazon p2.xlarge instance, cuda 9, cudnn7, no mkl ---

[GitHub] eric-haibin-lin opened a new pull request #10081: [MXNET-82] [WIP] Sparse op tutorial for developers

eric-haibin-lin opened a new pull request #10081: [MXNET-82] [WIP] Sparse op tutorial for developers URL: https://github.com/apache/incubator-mxnet/pull/10081 ## Description ## (Brief description on what this PR is about) ## Checklist ## ### Essentials ### - [ ] Passed code

[GitHub] KellenSunderland closed pull request #10080: WIP: Test building without explicit MKL disable

KellenSunderland closed pull request #10080: WIP: Test building without explicit MKL disable URL: https://github.com/apache/incubator-mxnet/pull/10080 This is a PR merged from a forked repository. As GitHub hides the original diff on merge, it is displayed below for the sake of provenance

[GitHub] kenfehling commented on issue #2910: Helper function to get GPU device number?

kenfehling commented on issue #2910: Helper function to get GPU device number? URL: https://github.com/apache/incubator-mxnet/issues/2910#issuecomment-372485992 When I try ```os.environ["CUDA_VISIBLE_DEVICES"]``` I get: ``` KeyError: 'CUDA_VISIBLE_DEVICES' ```

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372486000 Can you please supply build parameters that you use? ie CUDA, yes? CUDNN? MKL? etc.

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372485696 Modified script per your second code comment (smaller test set): ```python import mx

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372485876 I get the following (one GPU): [Epoch 0] accuracy=0.10 [Epoch 0] accuracy=0.1000

[GitHub] cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results

cjolivier01 commented on issue #9410: Training with the same parameters and seed gets significantly different results URL: https://github.com/apache/incubator-mxnet/issues/9410#issuecomment-372485413 If I run the script by itself, without the second training run, I get the same result eve

[GitHub] rahul003 commented on issue #9774: mx.io.ImageRecordIter does not respect dtype argument / FP16 performance on Volta

rahul003 commented on issue #9774: mx.io.ImageRecordIter does not respect dtype argument / FP16 performance on Volta URL: https://github.com/apache/incubator-mxnet/issues/9774#issuecomment-371654013 Both suggestions didn't help improve the speed unforunately. Using MXNET_CUDNN_AUTOTUNE_DE

[GitHub] rahul003 commented on issue #9774: mx.io.ImageRecordIter does not respect dtype argument / FP16 performance on Volta

rahul003 commented on issue #9774: mx.io.ImageRecordIter does not respect dtype argument / FP16 performance on Volta URL: https://github.com/apache/incubator-mxnet/issues/9774#issuecomment-372480640 Sorry I was digressing from the topic of the issue. Regarding the iterator issue, we need

[GitHub] rahul003 opened a new issue #9774: mx.io.ImageRecordIter does not respect dtype argument / FP16 performance on Volta

rahul003 opened a new issue #9774: mx.io.ImageRecordIter does not respect dtype argument / FP16 performance on Volta URL: https://github.com/apache/incubator-mxnet/issues/9774 ## Description `mx.io.ImageRecordIter` or `src/io/iter_image_recordio_2.cc` doesn't respect dtype parameter tak

[incubator-mxnet] branch nlp_toolkit updated: fix CorpusReader (#10079)

This is an automated email from the ASF dual-hosted git repository. zhasheng pushed a commit to branch nlp_toolkit in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git The following commit(s) were added to refs/heads/nlp_toolkit by this push: new 7e59d74 fix CorpusReader (

[GitHub] szha closed pull request #10079: Fix CorpusReader

szha closed pull request #10079: Fix CorpusReader URL: https://github.com/apache/incubator-mxnet/pull/10079 This is a PR merged from a forked repository. As GitHub hides the original diff on merge, it is displayed below for the sake of provenance: As this is a foreign pull request (from a

[GitHub] nswamy commented on issue #10063: Documentation API or file

nswamy commented on issue #10063: Documentation API or file URL: https://github.com/apache/incubator-mxnet/issues/10063#issuecomment-372477419 @lutzroeder We manage and answer user questions/how to's on https://discuss.mxnet.io/, please post this question there. I will close this issue

[GitHub] nswamy closed issue #10063: Documentation API or file

nswamy closed issue #10063: Documentation API or file URL: https://github.com/apache/incubator-mxnet/issues/10063 This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and u

[GitHub] nswamy commented on issue #10053: Get all labels from mx.io.ImageRecordIter

nswamy commented on issue #10053: Get all labels from mx.io.ImageRecordIter URL: https://github.com/apache/incubator-mxnet/issues/10053#issuecomment-372476272 We manage and answer user questions/how to's on https://discuss.mxnet.io/, please post this question there. I will close this issue

[GitHub] nswamy closed issue #10053: Get all labels from mx.io.ImageRecordIter

nswamy closed issue #10053: Get all labels from mx.io.ImageRecordIter URL: https://github.com/apache/incubator-mxnet/issues/10053 This is an automated message from the Apache Git Service. To respond to the message, please log

[GitHub] nswamy closed issue #10057: Multiple workers on single CPU training

nswamy closed issue #10057: Multiple workers on single CPU training URL: https://github.com/apache/incubator-mxnet/issues/10057 This is an automated message from the Apache Git Service. To respond to the message, please log o

[GitHub] nswamy commented on issue #10057: Multiple workers on single CPU training

nswamy commented on issue #10057: Multiple workers on single CPU training URL: https://github.com/apache/incubator-mxnet/issues/10057#issuecomment-372476119 We manage and answer user questions/how to's on https://discuss.mxnet.io/, please post this question there. I will close this issue.

[GitHub] Jerryzcn commented on issue #10042: Gluon dataloader crash on speech recognition training

Jerryzcn commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372471802 Here is the code that get stuck. Change num_worker=0 will work, however, on Mac with 1.0.0post3, this is not an issu

[GitHub] KellenSunderland opened a new pull request #10080: WIP: Test building without explicit MKL disable

KellenSunderland opened a new pull request #10080: WIP: Test building without explicit MKL disable URL: https://github.com/apache/incubator-mxnet/pull/10080 Testing CI when MKL isn't explicitly disabled. Temporary PR, will be closed in the future without merge. --

[GitHub] Jerryzcn commented on issue #10042: Gluon dataloader crash on speech recognition training

Jerryzcn commented on issue #10042: Gluon dataloader crash on speech recognition training URL: https://github.com/apache/incubator-mxnet/issues/10042#issuecomment-372471802 Here is the code that get stuck. Change num_worker=0 will work, however, on Mac, this is not an issue ``` from

[GitHub] KellenSunderland commented on a change in pull request #10075: Fix CMake build issue with MKL.

KellenSunderland commented on a change in pull request #10075: Fix CMake build issue with MKL. URL: https://github.com/apache/incubator-mxnet/pull/10075#discussion_r173954308 ## File path: ci/docker/runtime_functions.sh ## @@ -278,8 +278,6 @@ build_ubuntu_gpu_cmake() {

[GitHub] nswamy commented on issue #10060: Does mxnet.image suppor s3?

nswamy commented on issue #10060: Does mxnet.image suppor s3? URL: https://github.com/apache/incubator-mxnet/issues/10060#issuecomment-372468469 We manage and answer user questions/how to's on https://discuss.mxnet.io/, please post this question there. I will close this issue. ---

[GitHub] nswamy closed issue #10060: Does mxnet.image suppor s3?

nswamy closed issue #10060: Does mxnet.image suppor s3? URL: https://github.com/apache/incubator-mxnet/issues/10060 This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and

[GitHub] nswamy commented on issue #10063: Documentation API or file

nswamy commented on issue #10063: Documentation API or file URL: https://github.com/apache/incubator-mxnet/issues/10063#issuecomment-372468259 We don't have all in one file, but are generated into md files here: https://github.com/nswamy/incubator-mxnet/tree/master/docs/api/python ---

[GitHub] nswamy commented on issue #10066: WarpCTC loss output

nswamy commented on issue #10066: WarpCTC loss output URL: https://github.com/apache/incubator-mxnet/issues/10066#issuecomment-372467431 We manage and answer user questions/how to's on https://discuss.mxnet.io/, please post this question there. I will close this issue. ---

[GitHub] nswamy closed issue #10066: WarpCTC loss output

nswamy closed issue #10066: WarpCTC loss output URL: https://github.com/apache/incubator-mxnet/issues/10066 This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the

[GitHub] marcoabreu commented on a change in pull request #10075: Fix CMake build issue with MKL.

marcoabreu commented on a change in pull request #10075: Fix CMake build issue with MKL. URL: https://github.com/apache/incubator-mxnet/pull/10075#discussion_r173946287 ## File path: ci/docker/runtime_functions.sh ## @@ -278,8 +278,6 @@ build_ubuntu_gpu_cmake() { cmak

[GitHub] nswamy closed issue #10070: Distributed Training (Permission denied)

nswamy closed issue #10070: Distributed Training (Permission denied) URL: https://github.com/apache/incubator-mxnet/issues/10070 This is an automated message from the Apache Git Service. To respond to the message, please log

[GitHub] nswamy commented on issue #10070: Distributed Training (Permission denied)

nswamy commented on issue #10070: Distributed Training (Permission denied) URL: https://github.com/apache/incubator-mxnet/issues/10070#issuecomment-372461628 We manage and answer user questions/how to's on https://discuss.mxnet.io/, please post this question there. I will close this issue.

[GitHub] cjolivier01 commented on a change in pull request #10078: Support float16 in L2Normalization operator

cjolivier01 commented on a change in pull request #10078: Support float16 in L2Normalization operator URL: https://github.com/apache/incubator-mxnet/pull/10078#discussion_r173945220 ## File path: src/operator/l2_normalization.cc ## @@ -26,13 +26,18 @@ namespace mxnet { n

[GitHub] KellenSunderland commented on a change in pull request #10075: Fix CMake build issue with MKL.

KellenSunderland commented on a change in pull request #10075: Fix CMake build issue with MKL. URL: https://github.com/apache/incubator-mxnet/pull/10075#discussion_r173944839 ## File path: ci/docker/runtime_functions.sh ## @@ -278,8 +278,6 @@ build_ubuntu_gpu_cmake() {

[GitHub] cjolivier01 commented on a change in pull request #10078: Support float16 in L2Normalization operator

cjolivier01 commented on a change in pull request #10078: Support float16 in L2Normalization operator URL: https://github.com/apache/incubator-mxnet/pull/10078#discussion_r173944782 ## File path: src/operator/l2_normalization.cc ## @@ -26,13 +26,18 @@ namespace mxnet { n

[GitHub] nswamy commented on issue #10077: Fine tuning network

nswamy commented on issue #10077: Fine tuning network URL: https://github.com/apache/incubator-mxnet/issues/10077#issuecomment-372460019 We manage user-questions on https://discuss.mxnet.io/, please post your question there. I am closing this issue. ---

[GitHub] nswamy closed issue #10077: Fine tuning network

nswamy closed issue #10077: Fine tuning network URL: https://github.com/apache/incubator-mxnet/issues/10077 This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the

[GitHub] haojin2 commented on a change in pull request #10078: Support float16 in L2Normalization operator

haojin2 commented on a change in pull request #10078: Support float16 in L2Normalization operator URL: https://github.com/apache/incubator-mxnet/pull/10078#discussion_r173943169 ## File path: src/operator/l2_normalization.cc ## @@ -26,13 +26,18 @@ namespace mxnet { names

[GitHub] sxjscience opened a new pull request #10079: Fix CorpusReader

sxjscience opened a new pull request #10079: Fix CorpusReader URL: https://github.com/apache/incubator-mxnet/pull/10079 ## Description ## Fix the corpus reader @szha This is an automated message from the Apache Git Serv

  1   2   >