remove the hack #129
Merged
remove the hack #129
Conversation
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
It looks like we no longer need the hack..
aviatesk
commented
Mar 28, 2023
| @@ -226,8 +226,7 @@ z45, delta45 = frule_via_ad(DiffractorRuleConfig(), (0,1), x -> log(exp(x)), 2) | |||
| @test gradient(x -> sum(sqrt.(atan.(x, transpose(x)))), [1,2,3])[1] ≈ [0.2338, -0.0177, -0.0661] atol=1e-3 | |||
| @test gradient(x -> sum(exp.(log.(x))), [1,2,3]) == ([1,1,1],) | |||
|
|
|||
| # XXX the world-age limitation is preventing this test from passing | |||
| # @test gradient(x -> sum((exp∘log).(x)), [1,2,3]) == ([1,1,1],) # frule_via_ad | |||
| @test_broken gradient(x -> sum((exp∘log).(x)), [1,2,3]) == ([1,1,1],) # frule_via_ad | |||
There was a problem hiding this comment.
This case causes BoundsError now, which makes me doubt this is related to world-age anymore.
julia> using Diffractor
julia> gradient(x -> sum((exp∘log).(x)), [1,2,3])
ERROR: BoundsError: attempt to access Tuple{Diffractor.TangentBundle{1, ComposedFunction{typeof(exp), typeof(log)}, Diffractor.ExplicitTangent{Tuple{ChainRulesCore.NoTangent}}}, Diffractor.TangentBundle{1, Int64, Diffractor.ExplicitTangent{Tuple{Int64}}}} at index [3]
Stacktrace:
[1] (::Diffractor.∂☆recurse{1})(::Diffractor.TangentBundle{1, ComposedFunction{typeof(exp), typeof(log)}, Diffractor.ExplicitTangent{Tuple{ChainRulesCore.NoTangent}}}, ::Diffractor.TangentBundle{1, Int64, Diffractor.ExplicitTangent{Tuple{Int64}}})
@ Diffractor ~/julia/packages/Diffractor/src/stage1/recurse_fwd.jl:53
[2] (::Diffractor.∂☆internal{1})(::Diffractor.TangentBundle{1, ComposedFunction{typeof(exp), typeof(log)}, Diffractor.ExplicitTangent{Tuple{ChainRulesCore.NoTangent}}}, ::Vararg{Diffractor.AbstractTangentBundle{1}})
@ Diffractor ~/julia/packages/Diffractor/src/stage1/forward.jl:112
[3] frule_via_ad(::Diffractor.DiffractorRuleConfig, ::Tuple{ChainRulesCore.NoTangent, Int64}, ::Function, ::Vararg{Any})
@ Diffractor ~/julia/packages/Diffractor/src/stage1/forward.jl:120
[4] (::ChainRules.var"#1733#1735"{typeof(ChainRulesCore.frule_via_ad), Diffractor.DiffractorRuleConfig, ComposedFunction{typeof(exp), typeof(log)}})(a::Int64)
@ ChainRules ~/.julia/packages/ChainRules/RZYEu/src/rulesets/Base/broadcast.jl:109
[5] _broadcast_getindex_evalf
@ ./broadcast.jl:697 [inlined]
[6] _broadcast_getindex
@ ./broadcast.jl:670 [inlined]
[7] getindex
@ ./broadcast.jl:624 [inlined]
[8] iterate
@ ./broadcast.jl:261 [inlined]
[9] iterate
@ ./broadcast.jl:255 [inlined]
[10] #collect_structarray#94
@ ~/.julia/packages/StructArrays/F5fDf/src/collect.jl:39 [inlined]
[11] collect_structarray
@ ~/.julia/packages/StructArrays/F5fDf/src/collect.jl:37 [inlined]
[12] #StructArray#29
@ ~/.julia/packages/StructArrays/F5fDf/src/structarray.jl:255 [inlined]
[13] StructArray
@ ~/.julia/packages/StructArrays/F5fDf/src/structarray.jl:254 [inlined]
[14] unzip_broadcast
@ ~/.julia/packages/ChainRules/RZYEu/src/unzipped.jl:39 [inlined]
[15] split_bc_inner
@ ~/.julia/packages/ChainRules/RZYEu/src/rulesets/Base/broadcast.jl:108 [inlined]
[16] split_bc_forwards
@ ~/.julia/packages/ChainRules/RZYEu/src/rulesets/Base/broadcast.jl:104 [inlined]
[17] rrule
@ ~/.julia/packages/ChainRules/RZYEu/src/rulesets/Base/broadcast.jl:41 [inlined]
[18] ∂⃖
@ ~/julia/packages/Diffractor/src/stage1/generated.jl:207 [inlined]
[19] ∂⃖
@ ~/julia/packages/Diffractor/src/stage1/generated.jl:372 [inlined]
[20] broadcasted
@ ./broadcast.jl:1325 [inlined]
[21] (::Diffractor.∂⃖recurse{1})(::typeof(Base.Broadcast.broadcasted), ::ComposedFunction{typeof(exp), typeof(log)}, ::Vector{Int64})
@ Diffractor ~/julia/packages/Diffractor/src/stage1/generated.jl:0
[22] (::∂⃖{1})(::typeof(Base.Broadcast.broadcasted), ::Function, ::Vararg{Any})
@ Diffractor ~/julia/packages/Diffractor/src/stage1/generated.jl:209
[23] #3
@ ./REPL[2]:1 [inlined]
[24] (::Diffractor.∂⃖recurse{1})(::var"#3#4", ::Vector{Int64})
@ Diffractor ~/julia/packages/Diffractor/src/stage1/generated.jl:0
[25] ∂⃖
@ ~/julia/packages/Diffractor/src/stage1/generated.jl:209 [inlined]
[26] ∂⃖
@ ~/julia/packages/Diffractor/src/interface.jl:25 [inlined]
[27] (::Diffractor.∇{var"#3#4"})(args::Vector{Int64})
@ Diffractor ~/julia/packages/Diffractor/src/interface.jl:123
[28] Diffractor.∇(::Function, ::Vector{Int64})
@ Diffractor ~/julia/packages/Diffractor/src/interface.jl:130
[29] top-level scope
@ REPL[2]:1
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## main #129 +/- ##
==========================================
+ Coverage 48.06% 48.51% +0.45%
==========================================
Files 27 27
Lines 2657 2659 +2
==========================================
+ Hits 1277 1290 +13
+ Misses 1380 1369 -11
... and 2 files with indirect coverage changes Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report in Codecov by Sentry. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
It looks like we no longer need the hack..