You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
Megvii Engine Team d60d028a40 feat(mge/device): enable to get cuda/cudnn/tensorrt version 3 years ago
..
amp fix(imperative/amp): add is_scalar check in elemwise and concat 3 years ago
autodiff fix(imperative/amp): fix format transformation for symbol trans 3 years ago
core fix(mge/tools): fix network_visualize for op without out shapes 2 years ago
data chore(license): remove all license in file header 3 years ago
distributed feat(mge/distributed): add support for batch send recv op 3 years ago
dtr chore(license): remove all license in file header 3 years ago
experimental chore(license): remove all license in file header 3 years ago
functional feat(mgb): supports value infer and empty input tensor in ElemwiseMultiType 2 years ago
hub chore(license): remove all license in file header 3 years ago
jit feat(serialization): add no_change_graph and version param whem dump model 3 years ago
module feat(imperative/amp): add set_format 3 years ago
optimizer fix(imperative): fix inplace operation of optim 2 years ago
quantization chore(license): remove all license in file header 3 years ago
random chore(license): remove all license in file header 3 years ago
tools fix(mge/tools): fix network_visualize for op without out shapes 2 years ago
traced_module fix(traced_module): fix trace_module function may raise error in finally scope 3 years ago
utils fix(mge/tools): fix module_stats for duplicated module 2 years ago
__init__.py refactor(mgb): check input when profiling 3 years ago
_multistream.py perf(mge): improve cross stream memory borrowing 3 years ago
config.py fix(mge): control compute_mode by context 3 years ago
conftest.py ci(pytest): fixture doctest namespace in pytest config 3 years ago
device.py feat(mge/device): enable to get cuda/cudnn/tensorrt version 2 years ago
logger.py chore(license): remove all license in file header 3 years ago
serialization.py chore(license): remove all license in file header 3 years ago
tensor.py feat(imperative/amp): add set_format 3 years ago

MegEngine 安装包中集成了使用 GPU 运行代码所需的 CUDA 环境,不用区分 CPU 和 GPU 版。 如果想要运行 GPU 程序,请确保机器本身配有 GPU 硬件设备并安装好驱动。 如果你想体验在云端 GPU 算力平台进行深度学习开发的感觉,欢迎访问 MegStudio 平台

C++ Cuda Python C SVG other