v0.9.6

NNlib v0.9.6

Diff since v0.9.5

Merged pull requests:

• Bump AMDGPU compat to 0.6 (#530) (@pxl-th)
• Simplify runtests.jl (#531) (@CarloLucibello)
• bump CUDA compat (#535) (@CarloLucibello)

v0.9.5

NNlib v0.9.5

Diff since v0.9.4

Merged pull requests:

• Add bias_act! (#457) (@mcabbott)
• Add a new workflow to trigger the benchmark workflow (#522) (@skyleaworlder)
• Fix BenchmarkTrigger.yml (#524) (@skyleaworlder)
• Fix BenchmarkTrigger.yml (#526) (@skyleaworlder)

v0.9.4

NNlib v0.9.4

Diff since v0.9.3

Closed issues:

• ReverseDiff cannot differentiate maxpool (#484)

Merged pull requests:

• fix #484 -- broaden pooling method signatures (#485) (@manuelbb-upb)
• cudnn complex convolution via gauss trick (#517) (@nikopj)

v0.9.3

NNlib v0.9.3

Diff since v0.9.2

Closed issues:

• ∇conv_filter_direct! nonzero beta + flipkernel bug (#518)

Merged pull requests:

• nonzero beta + flipkernel bugfix (#519) (@nikopj)

v0.9.2

NNlib v0.9.2

Diff since v0.9.0

Closed issues:

• pointer(CuArray) is not defined (#131)
• NNPACK convolution issue (#203)
• move issues from NNlibCUDA.jl (#495)

Merged pull requests:

• move from bad thread-local to task-local (#497) (@IanButterworth)
• backports v0.8.21 (#498) (@CarloLucibello)
• port batchnorm rrule from Flux (#499) (@CarloLucibello)
• Update AMDGPU compat to 0.5 (#521) (@pxl-th)

v0.8.21

NNlib v0.8.21

Diff since v0.8.20

Merged pull requests:

• Add (log)softmax benchmarks (#491) (@ToucheSir)
• make NNlibCUDA an extension (#492) (@CarloLucibello)
• Fix nthreads on 1.9 (failure in the presence of interactive threads) (#496) (@IanButterworth)
• backports v0.8.21 (#498) (@CarloLucibello)

v0.9.0

NNlib v0.9.0

Diff since v0.8.20

Merged pull requests:

• Add (log)softmax benchmarks (#491) (@ToucheSir)
• make NNlibCUDA an extension (#492) (@CarloLucibello)
• Fix nthreads on 1.9 (failure in the presence of interactive threads) (#496) (@IanButterworth)

v0.8.20

NNlib v0.8.20

Diff since v0.8.19

Closed issues:

• Bugged fallback in grouped Convs (#369)
• NNlibAMDGPUExt: UndefVarError: MIOpen not defined (#482)

Merged pull requests:

• Add hand written gelu derivative (#480) (@chengchingwen)
• [NNlibAMDGPUExt] Load MIOpen module only if it is available (#483) (@pxl-th)
• Use KernelAbstractions.jl for upsample kernels (#486) (@pxl-th)
• Use KernelAbstractions.jl for gather/scatter kernels (#487) (@pxl-th)
• [AMDGPU] Add dispatch path for FP16 batched mul (#488) (@pxl-th)

v0.8.19

NNlib v0.8.19

Diff since v0.8.18

Merged pull requests:

• Add AMDGPU extension (#470) (@pxl-th)
• Merge NNlibCUDA as subpackage (#471) (@ToucheSir)
• Add dropout & attention tests for AMDGPU (#472) (@pxl-th)
• Allow regular convolution for AMDGPU (#473) (@pxl-th)
• Add downstream tests for Lux.jl (#474) (@avik-pal)
• change AMDGPUExt to NNlibAMDGPUExt (#477) (@CarloLucibello)

v0.8.18

NNlib v0.8.18

Diff since v0.8.17

Closed issues:

• Unstable performance (#461)

Merged pull requests:

• Fix conv with groups when falling in direct backend (#468) (@gabrielpreviato)