8000 Commits · Qoboty/neon · GitHub

More Web Proxy on the site http://driver.im/

Commits

Commits on Feb 23, 2017

Update for v1.8.2
Jennifer Myers
committed

Commits on Feb 22, 2017

switch to new tag line for neon in README & docs
Jennifer Myers
committed

Commits on Feb 20, 2017

Shuffle dataset before splitting into subsets

anlthms
authored and
apark263
committed
Support greyscale videos

anlthms
authored and
apark263
committed
Fix error in documentation

anlthms
authored and
apark263
committed
Fix formatting of conv layer description

anlthms
authored and
apark263
committed

Commits on Feb 9, 2017

Change default stage depth for cifar_msra example

apark263
authored and
Jennifer Myers
committed
update eval_sick arg requirement

Yinyin Liu
authored and
Jennifer Myers
committed

Commits on Jan 26, 2017

Make the whale calls example stable (#384)

anlthms
authored and
ursk
committed

Commits on Jan 18, 2017

Update for v1.8.1
Jennifer Myers
committed

Commits on Jan 12, 2017

Nesterov momentum was updating velocities incorrectly

tyler-nervana
authored and
Jennifer Myers
committed
Update conv and deconv defaults

Hanlin Tang
authored and
Jennifer Myers
committed
add dilation to object dict

Hanlin Tang
authored and
Jennifer Myers
committed

Commits on Jan 7, 2017

Prevent gpu backend from ignoring non-zero slope in Rectlinclip
tyler-nervana
committed

Commits on Dec 28, 2016

Update for v1.8.0
Jennifer Myers
committed

Commits on Dec 27, 2016

update faster-rcnn

zach-nervana
authored and
Jennifer Myers
committed

Commits on Dec 23, 2016

Fix winograd wrapper code to parse and ignore dilation params
Stewart Hall
committed
allow specifying the number of time steps to evaluate during beam_search

zach-nervana
authored and
Jennifer Myers
committed

Commits on Dec 22, 2016

Add more test cases for dilated convolution

Stewart Hall
authored and
Jennifer Myers
committed
Change slice step size for dilated convolution brop

Stewart Hall
authored and
Jennifer Myers
committed
Handle corner cases in dilated conv (#354)

anlthms
authored and
Jennifer Myers
committed
Add dilated conv support to all sass direct conv kernels

Stewart Hall
authored and
Jennifer Myers
committed
Fix bug in cuda dilated convolution update kernel and raise unit test tolerances

Stewart Hall
authored and
Jennifer Myers
committed
Add unit tests for dilated convolution

anlthms
authored and
Jennifer Myers
committed
Add support for dilated conv in the CPU backend

anlthms
authored and
Jennifer Myers
committed
Add dilated convolution support to CUDA C conv kernel and conv layer

Stewart Hall
authored and
Jennifer Myers
committed
move start_index in beamsearch to Decoder from Seq2Seq and document its existence

zach-nervana
authored and
Jennifer Myers
committed
move init_state_list size check to before any fprop to reduce time to exception

zach-nervana
authored and
Jennifer Myers
committed
add helpful error messages; allow using start symbol to seed beamsearch

zach-nervana
authored and
Jennifer Myers
committed
add MultiMetric to allow wrapping Metric classes with an index so they can be used in models with multiple outputs (models which use the Tree container)

zach-nervana
authored and
Jennifer Myers
committed
fix bugs in MultiCost support; allow evaluating one Cost at a time

zach-nervana
authored and
Jennifer Myers
committed
fix documentation errors

ami-GS
authored and
Jennifer Myers
committed
Add community-contributed docker from @sofianhw
Jennifer Myers
committed
serialize both encoder and decoder shapes. on configure, check for and handle deserialized shapes

ursk
authored and
Jennifer Myers
committed

Commits on Dec 21, 2016

add more helpful error message when cpu tensor is reshaped to incompatible size

zach-nervana
authored and
Jennifer Myers
committed

0