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 67575d582c feat(mge/opr): add interpolate bilinear mode 4 years ago
..
adaptive_pooling chore(mge): update copyright years 5 years ago
add_update chore(mge): update copyright years 5 years ago
argmxx feat(mgb): add padding opr for megbrain 4 years ago
argsort chore(mge): update copyright years 5 years ago
batch_conv_bias refactor(mgb/dnn): add negative attribute for algo 5 years ago
batch_normalization feat(log): opt log 5 years ago
batched_matrix_mul refactor(mgb/dnn): add negative attribute for algo 5 years ago
check_has_inf feat(mgb/opr): add check_has_inf kernel and opr 4 years ago
checksum chore(mge): update copyright years 5 years ago
concat chore(mge): update copyright years 5 years ago
cond_take chore(mge): update copyright years 5 years ago
conv_bias refactor(mgb): make conv handle noncontiguous tensors 4 years ago
convolution feat(dnn/cuda): add convolution with i8 input and i4 output 4 years ago
convolution3d refactor(mgb): make convolution3D handle noncontiguous tensors 4 years ago
convpooling chore(mge): update copyright years 5 years ago
correlation feat(dnn/cuda): add correlation kernel 5 years ago
cumsum chore(mge): update copyright years 5 years ago
cvt_color chore(mge): update copyright years 5 years ago
dct chore(mge): update copyright years 5 years ago
deformable_conv refactor(mgb/dnn): add negative attribute for algo 5 years ago
deformable_ps_roi_pooling chore(mge): update copyright years 5 years ago
dot chore(mge): update copyright years 5 years ago
elemwise feat(mge/opr): add silu and gelu 4 years ago
elemwise_multi_type feat(dnn/cuda): add cuda elemwise int4 5 years ago
eye chore(mge): update copyright years 5 years ago
fake_quant chore(mge): update copyright years 5 years ago
fill feat(dnn): add fill kernel 4 years ago
flip chore(mge): update copyright years 5 years ago
gaussian_blur chore(mge): update copyright years 5 years ago
group_local feat(log): opt log 5 years ago
images2neibs feat(mge/module): add python wrapper for unfold 5 years ago
indexing_multi_axis_vec feat(log): opt log 5 years ago
indexing_one_hot feat(log): opt log 5 years ago
linspace chore(mge): update copyright years 5 years ago
local feat(log): opt log 5 years ago
local_share refactor(mgb/dnn): add negative attribute for algo 5 years ago
lrn chore(mge): update copyright years 5 years ago
lsq feat(mgb/opr): add lsq opr 4 years ago
mask_conv chore(mge): update copyright years 5 years ago
matrix_inverse chore(mge): update copyright years 5 years ago
matrix_mul fix(dnn): fixes for int4 5 years ago
max_tensor_diff chore(mge): update copyright years 5 years ago
mesh_indexing chore(mge): update copyright years 5 years ago
padding feat(mgb): add padding opr for megbrain 4 years ago
param_pack chore(mge): update copyright years 5 years ago
pooling feat(dnn): add deterministic max pooling 4 years ago
powc chore(mge): update copyright years 5 years ago
reduce feat(log): opt log 5 years ago
relayout chore(mge): update copyright years 5 years ago
relayout_format feat(dnn/cuda): support transforming layout between nchw and nchw64 when channel not aligned to 64 5 years ago
remap feat(log): opt log 5 years ago
repeat chore(mge): update copyright years 5 years ago
resize feat(mge/opr): add interpolate bilinear mode 4 years ago
rng feat(mge/random): add some random op and remove random/distrbution.py 4 years ago
roi_align chore(mge): update copyright years 5 years ago
roi_copy chore(mge): update copyright years 5 years ago
roi_pooling chore(mge): update copyright years 5 years ago
rotate chore(mge): update copyright years 5 years ago
separable_conv chore(mge): update copyright years 5 years ago
separable_filter feat(log): opt log 5 years ago
sleep feat(bazel/windows/xp/sp2/inference): implement inference on windows xp 4 years ago
sliding_window_transpose feat(mge): add SlidingWindowTranspose opr 4 years ago
split chore(mge): update copyright years 5 years ago
svd chore(mge): update copyright years 5 years ago
tensor_remap chore(mge): update copyright years 5 years ago
tile chore(mge): update copyright years 5 years ago
topk chore(mge): update copyright years 5 years ago
tqt fix(mgb/dnn): fix backward computation of tqt 5 years ago
transpose chore(mge): update copyright years 5 years ago
type_cvt chore(mge): update copyright years 5 years ago
warp_affine feat(log): opt log 5 years ago
warp_perspective feat(dnn/cuda): add nhwc 4bit warp perspective 5 years ago
handle.cpp feat(mgb): add padding opr for megbrain 4 years ago
handle.h feat(dnn): add deterministic max pooling 4 years ago
lowbit_utils.cpp feat(dnn/cuda): add conv u4xs4 sass kernel 5 years ago
lowbit_utils.h feat(dnn/cuda): add conv u4xs4 sass kernel 5 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