Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

[Lane 2 · kotlin-core] Recall metric

未关闭
  • 哪里不对？
#1,228 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
3/5
预计耗时
1-3 小时
新手友好度
55/100
Issue 类型
功能
描述清晰度
描述清楚
活跃度
冷清
技术栈
kotlin
领域
machine-learning

调研方向

First check the dependency on #1226, then read Accuracy.kt and ConfusionMatrixAccumulator. Implement Recall and its factory functions in Recall.kt, and add the listed cases in RecallTest.kt, following AccuracyTest.kt for tensor helpers. Run the specified API dump and JVM tests; done means the public API is included and all six cases pass.

由索引模型根据 Issue 内容生成。

描述

coding good first issue size:s skill:kotlin-core sub-issue training

Sub-issue of #1222 (Precision, Recall and F1Score metrics).

Lane: 2 · Kotlin core
Skill needed: Kotlin. Read Accuracy.kt and ConfusionMatrixAccumulator first — this task is mostly wiring.
Size: s (~2–3 h)
Blocked by: #1226 (accumulator + Averaging)

What to do

  1. Add public class Recall(averaging: Averaging = Averaging.MACRO, dim: Int = -1, threshold: Float? = null) : Metric
    in skainet-lang/skainet-lang-core/src/commonMain/kotlin/sk/ainet/lang/nn/metrics/Recall.kt:
    • override val name = "recall";
    • update(...) delegates to a private ConfusionMatrixAccumulator(dim, threshold);
    • compute() = reduce(acc.counts(), averaging) { tp, fp, fn -> if (tp + fn == 0L) 0.0 else tp.toDouble() / (tp + fn) };
    • reset() delegates.
      Mirror Accuracy.kt's KDoc style (what the metric means, the dim / threshold semantics, what averaging does).
  2. Add factory functions at the bottom of the file mirroring accuracy() / binaryAccuracy():
    public fun recall(averaging: Averaging = Averaging.MACRO, dim: Int = -1): Metric and
    public fun binaryRecall(threshold: Float = 0.5f): Metric.
  3. Tests in skainet-backends/skainet-backend-cpu/src/commonTest/kotlin/sk/ainet/exec/metrics/RecallTest.kt,
    copying AccuracyTest.kt's tensor helpers. Cover, with hand-computed expected values
    (see the parent issue and #1224 for the reference fixture pred [2,0,1,2], target [2,1,1,0]):
    • macro multi-class with Int32 index targets → 0.5;
    • micro multi-class → 0.5 (and assert it equals Accuracy on the same batch);
    • the same batch with soft (FP32 one-hot) targets gives the same result;
    • binary with threshold, 2-class logits;
    • accumulation across two update() calls equals one call on the concatenated batch;
    • zero-division: a class with TP+FP+FN = 0 contributes 0.0 (per #1224), and
      compute() on a fresh metric returns 0.0.
  4. ./gradlew :skainet-lang:skainet-lang-core:apiDump and
    ./gradlew :skainet-backends:skainet-backend-cpu:jvmTest --tests "sk.ainet.exec.metrics.*".

Acceptance

  • Recall + recall() + binaryRecall() public and in the API dump
  • All six test cases above pass on JVM
  • No duplicated argmax / iteration logic — everything goes through the accumulator
  • PR links this issue and #1222; PR body states "unit tests only; ground-truth wiring tracked in SKEEP-005 (#1223)"

Notes

Recall answers "of what was actually positive, how much did I catch". The twin task #1227 (Precision) is the same shape with tp / (tp + fp) — filed separately so two people can each ship one.

主要语言
Kotlin
星标
52
派生
15
平均合并
1 天 1 小时
30 天内合并 PR
41

环境准备

  • 没有 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

SKaiNET-developers/SKaiNET 的其他 Issue

查看 SKaiNET-developers/SKaiNET 的全部 Issue

相似的 Issue

更多 Kotlin Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。