first | « prev | next »
AlexeyAB
2018-08-21 f606b5456e4876da5f90e2902b2dff07516a03dc
2018-08-21 AlexeyAB
XNOR-net 21 FPS on CPU yolov2-tiny.cfg
blob@ f606b5 commitdiff | diff to current
2018-08-13 AlexeyAB
Some fixes for AVX support on CPU
blob@ f92b20 commitdiff | diff to current
2018-08-12 AlexeyAB
Fixed AVX compiled bug
blob@ b1dddf commitdiff | diff to current
2018-08-10 AlexeyAB
Experiments
blob@ 1f2155 commitdiff | diff to current
2018-08-10 AlexeyAB
Bug fixes. Tested im2col_cpu_custom_transpose - bad way.
blob@ a9fef1 commitdiff | diff to current
2018-08-09 AlexeyAB
Optimized: transpose
blob@ 3e856e commitdiff | diff to current
2018-08-08 AlexeyAB
Optimized on CPU: gemm_bin, im2col, activation, transpose
blob@ d6162a commitdiff | diff to current
2018-08-08 AlexeyAB
Try to use avx_hs() - slow and requires alignment 4096 bits < (l.size*l.siz...
blob@ a284a7 commitdiff | diff to current
2018-08-07 AlexeyAB
XNOR-net on CPU AVX2
blob@ 0a326e commitdiff | diff to current
2018-07-10 AlexeyAB
Just used spaces for indents instead of Tabs
blob@ cfc5fe commitdiff | diff to current
2018-05-23 AlexeyAB
Fixed memory leaks for Yolo: train, test
blob@ ec6883 commitdiff | diff to current
2018-05-19 AlexeyAB
Fixed xnor for random=1
blob@ c1bb8c commitdiff | diff to current
2018-05-14 AlexeyAB
Added BFLOPs output for network configurations
blob@ 8b5344 commitdiff | diff to current
2018-05-03 iovodov
Output improvements for detector results: When printing detector results, o...
blob@ 028696 commitdiff | diff to current
2018-04-16 AlexeyAB
Accelerated by another 5% using FP16/32 Batch-norm for Tensor Cores.
blob@ 9bae70 commitdiff | diff to current
2018-04-16 AlexeyAB
Loss-graph store automatically (iterations == max_batches) at the end of tr...
blob@ c52fa4 commitdiff | diff to current
2018-04-14 AlexeyAB
Fixed bug in Tensor Cores V100 (1. Desc in Batch norm, 2. Manually selected...
blob@ eb9c88 commitdiff | diff to current
2018-03-19 AlexeyAB
Improve training performance - batch-norm using cuDNN.
blob@ 537d13 commitdiff | diff to current
2018-03-09 AlexeyAB
Fixed multi-GPU training for Tensor Cores
blob@ 880cf1 commitdiff | diff to current
2018-02-25 AlexeyAB
Added support for Tensor Cores CC >= 7.0 (V100). For FP16/32 (mixed precisi...
blob@ cad4d1 commitdiff | diff to current
2018-02-23 AlexeyAB
Updated to CUDA 9.1. And fixed no_gpu dependecies.
blob@ cd2bde commitdiff | diff to current
2018-02-22 AlexeyAB
Fix
blob@ f558d5 commitdiff | diff to current
2018-02-22 AlexeyAB
Use half_float16 instead of float32 if defined both CUDNN and CUDNN_HALF. U...
blob@ dda993 commitdiff | diff to current
2018-02-21 AlexeyAB
If there is excessive GPU-RAM consumption by CUDNN then then do not use Wor...
blob@ 033e93 commitdiff | diff to current
2018-02-21 AlexeyAB
Optimized resizing of network for random=1
blob@ 4b0be8 commitdiff | diff to current
2017-08-02 AlexeyAB
cuDNN 6.0 supported. Also speed of console example improved.
blob@ bc8100 commitdiff | diff to current
2017-01-16 AlexeyAB
Fixed behavior if missing library cudnn.lib
blob@ d7a30a commitdiff | diff to current
2017-01-15 AlexeyAB
Fixed behavior if missing library cudnn.lib
blob@ 3b9afd commitdiff | diff to current
2016-11-19 Joseph Redmon
cpu batch norm works
blob@ 62235e commitdiff | diff to current
2016-11-16 Joseph Redmon
:fire: :fire: :dragonite:
blob@ fc9b86 commitdiff | diff to current
2016-11-16 Joseph Redmon
hey
blob@ 0d6b10 commitdiff | diff to current
2016-11-05 Joseph Redmon
new font strategy
blob@ c7a700 commitdiff | diff to current
2016-10-26 Joseph Redmon
ADAM
blob@ 352ae7 commitdiff | diff to current
2016-09-25 Joseph Redmon
So I have this new programming paradigm.......
blob@ 481b57 commitdiff | diff to current
2016-09-20 Joseph Redmon
better multigpu
blob@ 73f7aa commitdiff | diff to current
2016-09-12 Joseph Redmon
good chance I didn't break anything
blob@ 5c067d commitdiff | diff to current
2016-09-01 Joseph Redmon
updates and things
blob@ 8f1b4e commitdiff | diff to current
2016-08-05 Joseph Redmon
some more stuff
blob@ 845ab7 commitdiff | diff to current
2016-07-19 Joseph Redmon
updates
blob@ 936129 commitdiff | diff to current
2016-06-19 Joseph Redmon
no mean on input binarization
blob@ 08c7cf commitdiff | diff to current
2016-06-14 Joseph Redmon
hate warnings
blob@ 8322a5 commitdiff | diff to current
2016-06-08 Joseph Redmon
idk just in case
blob@ 752094 commitdiff | diff to current
2016-06-06 Joseph Redmon
stuff for carlo
blob@ 8a767f commitdiff | diff to current
2016-06-06 Joseph Redmon
tactics
blob@ 4625a1 commitdiff | diff to current
2016-06-02 Joseph Redmon
hope i didn't break anything
blob@ ec3d05 commitdiff | diff to current
2016-05-13 Joseph Redmon
fixed
blob@ 881d6e commitdiff | diff to current
2016-05-13 Joseph Redmon
art, cudnn
blob@ 13209d commitdiff | diff to current
2016-05-06 Joseph Redmon
so much need to commit
blob@ c7b10c commitdiff | diff to current
2016-03-16 Joseph Redmon
go updates
blob@ cff59b commitdiff | diff to current
2016-01-28 Joseph Redmon
lots of stuff
blob@ 913d35 commitdiff | diff to current
first | « prev | next »