autumnai/collenchyma-nn issues and pull requests

#48 - Refactor sync

Pull Request - State: open - Opened by alexandermorozov almost 9 years ago - 1 comment

#47 - docs/readme: add cuDNN v4 to feature matrix

Pull Request - State: closed - Opened by MichaelHirn almost 9 years ago - 3 comments

#46 - Refactor plugin impls to be generic; implement ndim convolution (no grad for now); fix conv tests.

Pull Request - State: open - Opened by DiamondLovesYou almost 9 years ago - 4 comments

#45 - Convolution tests fail.

Issue - State: open - Opened by DiamondLovesYou almost 9 years ago

#44 - mem::transmute: Compile error on 32bit linux

Issue - State: open - Opened by bklooste almost 9 years ago - 2 comments

#43 - Update Feature Matrix for cuDNN v4

Issue - State: closed - Opened by MichaelHirn almost 9 years ago

#42 - Tracking Collenchyma issues

Issue - State: open - Opened by hobofan almost 9 years ago

#41 - chore/release: prepare 0.3.4 release

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#40 - fix/convolution: workaround for 0 memory allocation

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#39 - chore/release: prepare 0.3.3 release

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#38 - Remove our workspace size checking

Issue - State: closed - Opened by hobofan almost 9 years ago - 1 comment
Labels: bug, enhancement

#37 - Feat/passive cudnnv4

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#36 - chore/release: prepare 0.3.2 release

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#35 - Feature/convolution workspace

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#34 - Make use of Collenchyma prelude module

Issue - State: closed - Opened by hobofan almost 9 years ago
Labels: chore

#33 - Add prelude module

Issue - State: open - Opened by hobofan almost 9 years ago
Labels: refactor

#32 - chore/version: prepare for 0.3.1 release

Pull Request - State: closed - Opened by MichaelHirn almost 9 years ago - 4 comments

#31 - feat/pointwise: add pointwise activation functions

Pull Request - State: closed - Opened by MichaelHirn almost 9 years ago - 6 comments
Labels: F-CUDA

#30 - chore/release: prepare 0.3.0 release

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#29 - chore/collenchyma: bump collenchyma to 0.0.8

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#28 - Add support for in-place activation functions

Issue - State: closed - Opened by hobofan almost 9 years ago - 1 comment
Labels: Perf-Memory, F-All

#27 - Allow seperate workspaces for convolution/pooling

Issue - State: open - Opened by hobofan almost 9 years ago
Labels: F-CUDA, Perf-Memory, breaking, refactor

#26 - feat/cuda: share workspace between CUDA convolution operations

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#25 - feat/nn_trait: remove trait bounds for NN

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#24 - feat/log_softmax: add LogSoftmax operations

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#23 - Remove Float boundary for INN trait

Issue - State: closed - Opened by hobofan almost 9 years ago
Labels: enhancement, E-easy

#22 - feat/activation: allow CUDA activation to work with 1D/2D tensors

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#21 - feat/softmax: use an override CUDNNTensorDescriptor for 1-3D softmax

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 3 comments

#20 - feature/softmax: use an override CUDNNTensorDescriptor for 1-3D softmax

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 5 comments

#19 - refactor/cudnn: refactor ICudnnDesc impl into macro

Pull Request - State: closed - Opened by hobofan almost 9 years ago - 4 comments

#18 - Share workspace for convolution operations

Issue - State: closed - Opened by hobofan about 9 years ago
Labels: enhancement, F-CUDA

#17 - convolutions and license

Pull Request - State: closed - Opened by hobofan about 9 years ago - 3 comments

#16 - Allow usage of 1- and 2-dimensional tensors in CUDA

Issue - State: closed - Opened by hobofan about 9 years ago - 1 comment
Labels: bug, enhancement, F-CUDA

#15 - add CHANGELOG

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago - 4 comments

#14 - Bump to 0.2.1

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago - 4 comments

#13 - fix/native: Fix sigmoid_grad to use x_diff instead of x for dx

Pull Request - State: closed - Opened by ehiggs about 9 years ago - 4 comments

#12 - Add support for softmax w/ test and benches.

Pull Request - State: closed - Opened by ehiggs about 9 years ago - 8 comments

#11 - Refactor implementation structure and plugin traits

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago

#10 - Relicense under dual MIT/Apache-2.0

Issue - State: closed - Opened by emberian about 9 years ago - 4 comments

#9 - parameter ordering is inconsistent between pooling and convolution

Issue - State: open - Opened by hobofan about 9 years ago
Labels: breaking, refactor

#8 - Add bench tests to track performance improvements

Issue - State: open - Opened by ehiggs about 9 years ago - 2 comments

#7 - feat/native Implemnt Sigmoid, ReLU, and tanh for Native back ends

Pull Request - State: closed - Opened by ehiggs about 9 years ago - 13 comments

#6 - better build flags and native stub

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago - 3 comments

#5 - Ehiggs native

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago

#4 - test/test Add function stubs for native and applicable (ignored) tests.

Pull Request - State: closed - Opened by ehiggs about 9 years ago - 1 comment

#3 - cuDNN implementation for 0.1.0

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago - 3 comments

#2 - feat/nn: add all cudnn available operations to collenchyma-nn

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago

#1 - Fixes, refactoring and first NN activation implementation for CUDA

Pull Request - State: closed - Opened by MichaelHirn about 9 years ago - 3 comments

GitHub / autumnai/collenchyma-nn issues and pull requests