FluxML/Zygote.jl

Unable to differentiate a broadcasted constructor with CUDA array inputs

Open

#1,528 建立於 2024年10月3日

在 GitHub 查看
 (2 留言) (0 反應) (0 負責人)Julia (1,568 star) (220 fork)batch import
CUDAhelp wanted

描述

Due to a CUDA bug: https://github.com/JuliaGPU/CUDA.jl/issues/2514, crossposted for reference.

Encountered when differentiating over:

using Distributions
using CUDA
a = cu(ones(5)); b = cu(zeros(5));

Normal.(a, b)

貢獻者指南

技術棧
無
領域
machine learning
議題類型
bug
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
4
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
3-5 days
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
blocked
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
clear
前置要求
Julia basicsZygote.jl basicsCUDA.jl basicsBroadcasting in Julia
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
20
研究方向
The issue is blocked by a CUDA.jl bug (reference: JuliaGPU/CUDA.jl#2514). To proceed, first understand that bug and its implications on Zygote's differentiation of broadcasted constructors. Then, examine Zygote's handling of broadcasted calls and see if a workaround is possible. The provided example uses Distributions.Normal with CUDA arrays; review the interaction between Distributions and Zygote. No linked PRs or assignees exist yet.
Unable to differentiate a broadcasted constructor with CUDA array inputs · FluxML/Zygote.jl#1528 | Good First Issue