JuliaDiff / JuliaDiff/DifferentiationInterface.jl
Implement `Cache` for Enzyme reverse mode
Open
Nobody has claimed this yet.
backend
- Dominant language
- Julia
- Stars
- 313
- Forks
- 35
- PR merge metrics
- No merged PRs in 30d
Description
At the moment this seems to break the REPL:
using DifferentiationInterface, DifferentiationInterfaceTest
import DifferentiationInterfaceTest as DIT
using Enzyme: Enzyme
scens = filter(
s -> (
DIT.function_place(s) == :out &&
DIT.operator_place(s) == :in &&
DIT.operator(s) == :pullback &&
s.x isa Vector &&
s.y isa Vector &&
length(s.x) == 1 &&
s.tang isa NTuple{2}
),
default_scenarios(;
include_normal=false, include_constantified=false, include_cachified=true
),
)
test_differentiation(AutoEnzyme(; mode=Enzyme.Reverse), scens; logging=true) # fails
After I run it, the behavior of Julia itself is altered:
julia> a = 1
ERROR: UndefVarError: `get_binding_type` not defined in `Core`
Suggestion: check for spelling errors or missing imports.
Stacktrace:
[1] top-level scope
@ ~/Documents/GitHub/Julia/DifferentiationInterface.jl/DifferentiationInterface/test/Back/Enzyme/playground.jl:21
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the Enzyme reverse-mode scenario in the reported test_differentiation call and the playground.jl location shown in the stack trace. Reproduce the failure with logging enabled, then verify that the cachified scenario succeeds without leaving Julia's REPL unable to resolve get_binding_type.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- julia
- Domain
- backend
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100