This repository has been archived by the owner on Nov 4, 2024. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 1
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #25 from LuxDL/ap/remove_pt
Cleaning up of the codebase
- Loading branch information
Showing
20 changed files
with
495 additions
and
595 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file was deleted.
Oops, something went wrong.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,19 @@ | ||
| name: Code Quality Check | ||
|
|
||
| on: [pull_request] | ||
|
|
||
| jobs: | ||
| code-style: | ||
| name: Format Suggestions | ||
| runs-on: ubuntu-latest | ||
| steps: | ||
| - uses: julia-actions/julia-format@v3 | ||
|
|
||
| typos-check: | ||
| name: Spell Check with Typos | ||
| runs-on: ubuntu-latest | ||
| steps: | ||
| - name: Checkout Actions Repository | ||
| uses: actions/checkout@v4 | ||
| - name: Check spelling | ||
| uses: crate-ci/[email protected] |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,2 @@ | ||
| [default.extend-words] | ||
| nin = "nin" |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,13 +1,13 @@ | ||
| name = "WeightInitializers" | ||
| uuid = "d49dbf32-c5c2-4618-8acc-27bb2598ef2d" | ||
| authors = ["Avik Pal <[email protected]> and contributors"] | ||
| version = "0.1.7" | ||
| version = "0.1.8" | ||
|
|
||
| [deps] | ||
| ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4" | ||
| GPUArraysCore = "46192b85-c4d5-4398-a991-12ede77f4527" | ||
| LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e" | ||
| PartialFunctions = "570af359-4316-4cb7-8c74-252c00c2016b" | ||
| PrecompileTools = "aea7be01-6a6a-4083-8856-8a6e6704d82a" | ||
| Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c" | ||
| SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b" | ||
| Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2" | ||
|
|
@@ -19,26 +19,30 @@ CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba" | |
| WeightInitializersCUDAExt = "CUDA" | ||
|
|
||
| [compat] | ||
| Aqua = "0.8" | ||
| CUDA = "5" | ||
| ChainRulesCore = "1.21" | ||
| LinearAlgebra = "1.9" | ||
| Aqua = "0.8.7" | ||
| CUDA = "5.3.2" | ||
| ChainRulesCore = "1.23" | ||
| Documenter = "1.5.0" | ||
| ExplicitImports = "1.6.0" | ||
| GPUArraysCore = "0.1.6" | ||
| LinearAlgebra = "1.10" | ||
| PartialFunctions = "1.2" | ||
| PrecompileTools = "1.2" | ||
| Random = "1.9" | ||
| Random = "1.10" | ||
| ReTestItems = "1.24.0" | ||
| SpecialFunctions = "2" | ||
| StableRNGs = "1" | ||
| Statistics = "1.9" | ||
| Test = "1.9" | ||
| julia = "1.9" | ||
| Statistics = "1.10" | ||
| Test = "1.10" | ||
| julia = "1.10" | ||
|
|
||
| [extras] | ||
| Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595" | ||
| CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba" | ||
| Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c" | ||
| Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4" | ||
| ExplicitImports = "7d51a73a-1435-4ff3-83d9-f097790105c7" | ||
| ReTestItems = "817f1d60-ba6b-4fd5-9520-3cf149f6a823" | ||
| StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3" | ||
| Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2" | ||
| Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40" | ||
|
|
||
| [targets] | ||
| test = ["Aqua", "Test", "StableRNGs", "Random", "Statistics", "CUDA"] | ||
| test = ["Aqua", "CUDA", "Documenter", "ExplicitImports", "ReTestItems", "StableRNGs", "Test"] | ||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,79 +1,15 @@ | ||
| module WeightInitializersCUDAExt | ||
|
|
||
| using WeightInitializers, CUDA | ||
| using Random | ||
| import WeightInitializers: __partial_apply, NUM_TO_FPOINT, identity_init, sparse_init, | ||
| orthogonal | ||
| using CUDA: CUDA, CURAND | ||
| using WeightInitializers: WeightInitializers | ||
|
|
||
| const AbstractCuRNG = Union{CUDA.RNG, CURAND.RNG} | ||
|
|
||
| for T in ("16", "32", "64", "C16", "C32", "C64"), fname in (:ones, :zeros) | ||
| name = Symbol(fname, T) | ||
| TP = NUM_TO_FPOINT[Symbol(T)] | ||
| @eval begin | ||
| function WeightInitializers.$(name)(rng::AbstractCuRNG, dims::Integer...; kwargs...) | ||
| return CUDA.$(fname)($TP, dims...; kwargs...) | ||
| end | ||
| end | ||
|
|
||
| @eval function WeightInitializers.$(name)(rng::AbstractCuRNG; kwargs...) | ||
| return __partial_apply($name, (rng, (; kwargs...))) | ||
| end | ||
| end | ||
|
|
||
| function sparse_init(rng::AbstractCuRNG, ::Type{T}, dims::Integer...; | ||
| sparsity::Number, std::Number=T(0.01)) where {T <: Number} | ||
| if length(dims) != 2 | ||
| throw(ArgumentError("Only 2-dimensional outputs are supported for sparse initialization.")) | ||
| end | ||
|
|
||
| rows, cols = dims | ||
| prop_zero = min(1.0, sparsity) | ||
| num_zeros = ceil(Integer, prop_zero * rows) | ||
| sparse_array = randn(rng, T, dims...) .* T(std) | ||
| sparse_array[1:num_zeros, :] .= CUDA.zero(T) | ||
|
|
||
| return CUDA.@allowscalar mapslices(shuffle, sparse_array, dims=1) | ||
| function WeightInitializers.__zeros(::AbstractCuRNG, T::Type, dims::Integer...) | ||
| return CUDA.zeros(T, dims...) | ||
| end | ||
|
|
||
| function identity_init(rng::AbstractCuRNG, ::Type{T}, dims::Integer...; | ||
| gain::Number=1, shift::Integer=0) where {T <: Number} | ||
| if length(dims) == 1 | ||
| # Bias initialization | ||
| return CUDA.zeros(T, dims...) | ||
| elseif length(dims) == 2 | ||
| # Matrix multiplication | ||
| rows, cols = dims | ||
| mat = CUDA.zeros(T, rows, cols) | ||
| diag_indices = 1:min(rows, cols) | ||
| CUDA.fill!(view(mat, diag_indices, diag_indices), T(gain)) | ||
| return CUDA.circshift(mat, shift) | ||
| else | ||
| # Convolution or more dimensions | ||
| nin, nout = dims[end - 1], dims[end] | ||
| centers = map(d -> cld(d, 2), dims[1:(end - 2)]) | ||
| weights = CUDA.zeros(T, dims...) | ||
| #we should really find a better way to do this | ||
| CUDA.@allowscalar for i in 1:min(nin, nout) | ||
| index = (centers..., i, i) | ||
| weights[index...] = T(gain) | ||
| end | ||
| return CUDA.circshift(weights, (ntuple(d -> 0, length(dims) - 2)..., shift, shift)) | ||
| end | ||
| end | ||
|
|
||
| for initializer in (:sparse_init, :identity_init) | ||
| @eval function ($initializer)(rng::AbstractCuRNG, dims::Integer...; kwargs...) | ||
| return $initializer(rng, Float32, dims...; kwargs...) | ||
| end | ||
|
|
||
| @eval function ($initializer)(rng::AbstractCuRNG; kwargs...) | ||
| return __partial_apply($initializer, (rng, (; kwargs...))) | ||
| end | ||
| @eval function ($initializer)(rng::AbstractCuRNG, | ||
| ::Type{T}; kwargs...) where {T <: Number} | ||
| return __partial_apply($initializer, ((rng, T), (; kwargs...))) | ||
| end | ||
| function WeightInitializers.__ones(::AbstractCuRNG, T::Type, dims::Integer...) | ||
| return CUDA.ones(T, dims...) | ||
| end | ||
|
|
||
| end |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,62 +1,27 @@ | ||
| module WeightInitializers | ||
|
|
||
| import PrecompileTools: @recompile_invalidations | ||
| #! format: off | ||
| using ChainRulesCore: ChainRulesCore | ||
| using GPUArraysCore: @allowscalar | ||
| using LinearAlgebra: LinearAlgebra, Diagonal, qr | ||
| using PartialFunctions: :$ | ||
| using Random: Random, AbstractRNG, Xoshiro, shuffle | ||
| using SpecialFunctions: SpecialFunctions, erf, erfinv | ||
| using Statistics: Statistics, std | ||
| #! format: on | ||
|
|
||
| @recompile_invalidations begin | ||
| using ChainRulesCore, PartialFunctions, Random, SpecialFunctions, Statistics, | ||
| LinearAlgebra | ||
| end | ||
| const CRC = ChainRulesCore | ||
|
|
||
| include("utils.jl") | ||
| include("initializers.jl") | ||
|
|
||
| # Mark the functions as non-differentiable | ||
| for f in [ | ||
| :zeros64, | ||
| :ones64, | ||
| :rand64, | ||
| :randn64, | ||
| :zeros32, | ||
| :ones32, | ||
| :rand32, | ||
| :randn32, | ||
| :zeros16, | ||
| :ones16, | ||
| :rand16, | ||
| :randn16, | ||
| :zerosC64, | ||
| :onesC64, | ||
| :randC64, | ||
| :randnC64, | ||
| :zerosC32, | ||
| :onesC32, | ||
| :randC32, | ||
| :randnC32, | ||
| :zerosC16, | ||
| :onesC16, | ||
| :randC16, | ||
| :randnC16, | ||
| :glorot_normal, | ||
| :glorot_uniform, | ||
| :kaiming_normal, | ||
| :kaiming_uniform, | ||
| :truncated_normal, | ||
| :orthogonal, | ||
| :sparse_init, | ||
| :identity_init | ||
| ] | ||
| @eval @non_differentiable $(f)(::Any...) | ||
| end | ||
| include("autodiff.jl") | ||
|
|
||
| export zeros64, ones64, rand64, randn64, zeros32, ones32, rand32, randn32, zeros16, ones16, | ||
| rand16, randn16 | ||
| export zerosC64, onesC64, randC64, randnC64, zerosC32, onesC32, randC32, randnC32, zerosC16, | ||
| onesC16, randC16, randnC16 | ||
| export glorot_normal, glorot_uniform | ||
| export kaiming_normal, kaiming_uniform | ||
| export truncated_normal | ||
| export orthogonal | ||
| export sparse_init | ||
| export identity_init | ||
| export truncated_normal, orthogonal, sparse_init, identity_init | ||
|
|
||
| end |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,8 @@ | ||
| # Mark the functions as non-differentiable | ||
| for f in [:zeros64, :ones64, :rand64, :randn64, :zeros32, :ones32, :rand32, :randn32, | ||
| :zeros16, :ones16, :rand16, :randn16, :zerosC64, :onesC64, :randC64, | ||
| :randnC64, :zerosC32, :onesC32, :randC32, :randnC32, :zerosC16, :onesC16, | ||
| :randC16, :randnC16, :glorot_normal, :glorot_uniform, :kaiming_normal, | ||
| :kaiming_uniform, :truncated_normal, :orthogonal, :sparse_init, :identity_init] | ||
| @eval CRC.@non_differentiable $(f)(::Any...) | ||
| end |
Oops, something went wrong.
ed1f825There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@JuliaRegistrator register
ed1f825There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Registration pull request created: JuliaRegistries/General/109917
Tip: Release Notes
Did you know you can add release notes too? Just add markdown formatted text underneath the comment after the text
"Release notes:" and it will be added to the registry PR, and if TagBot is installed it will also be added to the
release that TagBot creates. i.e.
To add them here just re-invoke and the PR will be updated.
Tagging
After the above pull request is merged, it is recommended that a tag is created on this repository for the registered package version.
This will be done automatically if the Julia TagBot GitHub Action is installed, or can be done manually through the github interface, or via: