ml-explore/mlx

[BUG] Missing `array-api` methods

Open

#3,484 建立於 2026年5月5日

在 GitHub 查看
 (0 留言) (2 反應) (0 負責人)C++ (26,216 star) (1,794 fork)batch import
enhancementgood first issuepython

描述

Describe the bug The current array-api implementation is currently missing a few methods

To Reproduce

from_dlpack:

import mlx.core as mx

mx.array([[0, 1], [2, 3]]).__array_namespace__().from_dlpack

as_array(..., copy=...):

import mlx.core as mx

mx.array([[0, 1], [2, 3]]).__array_namespace__().asarray(mx.array([[0, 1], [2, 3]]), copy=True)

Expected behavior I would expect the above to work. I assume there are others as well.

Desktop (please complete the following information):

  • OS Version: 15.1 (24B83)
  • Version: 0.31.2

Additional context

貢獻者指南

技術棧
pythoncpp
領域
backenddata
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。
3
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。
half day
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。
fresh
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
Basic PythonFamiliarity with MLXUnderstanding of array api standard
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。
80
研究方向
Identify the source files implementing the array api namespace, likely in `mlx/core` or `mlx/python`. Look for existing methods like ` array namespace ` and `from dlpack` or `asarray`. Implement the missing `from dlpack` and `asarray` (with `copy` parameter) methods following the array api standard. Test with the provided reproduction code to ensure they work as expected.