GitHub · Where software is built

This repository was archived by the owner on Nov 17, 2023. It is now read-only.

[Development] MXNet 2.0 Update
#18931 · szha opened on Aug 15, 2020
19
[RFC] MXNet 2.0 JVM Language development
#17783 · lanking520 opened on Mar 6, 2020
55
[RFC] Future of Apache MXNet
#21206 · szha opened on Jun 21, 2023
5

Labels Milestones

GPU computation failing with CUDA Driver: an illegal memory access was encountered (Linux)

#19841

· matteosal opened

on Feb 4, 2021

Results are significant different between RTX 2080Ti and RTX 3090

#19649

· chinakook opened

on Dec 9, 2020

CTC loss in mxnet 2.0 not work when use half precision

#19628

· sdli1995 opened

on Dec 4, 2020

Multiple gpu train error

#19577

· chinakook opened

on Nov 23, 2020

CUDA: an illegal memory access was encountered on hybridized yolo model

#18834

· stu1130 opened

on Jul 31, 2020

gluon.nn.BatchNorm seems to swap updated values of moving_mean and moving_var on GPU.

#18751

· gilbertfrancois opened

on Jul 17, 2020

batch_dot precision issue in fp16

#18584

· pierre-cordani-nerd opened

on Jun 18, 2020

Failed test: test_gluon_gpu.test_rnn_unroll_variant_length

#16725

· anexpn opened

on Nov 5, 2019

Docker images are using CUDNN5 and CUDA8

#16332

· eric-haibin-lin opened

on Sep 30, 2019

Saving and loading cudNN autotune and graph optimization

Call for Contribution

Feature request

#16173

· QueensGambit opened

on Sep 14, 2019

__CUDACC_VER_ is no longer support

Pending Requester Info

#15913

· linhanxiao opened

on Aug 15, 2019

Convolution may have some precision problem with autotuned cudnn

#15638

· xziya opened

on Jul 23, 2019