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 6 · review] tokenizer.json parity — DARC review

未关闭
#1,330 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
4/5
预计耗时
1-2 天
新手友好度
18/100
Issue 类型
功能
描述清晰度
描述清楚
活跃度
活跃
技术栈
kotlin, python

调研方向

Read parent issue #1321 and compare each acceptance criterion with the shipped tokenizer code and docs. Regenerate goldens with the pinned tokenizers version using the Lane 4 script, then run the listed Gradle fixture and test tasks and hand-check three cases in Python. Verify the cited semantics and the specified failure behavior, added-token vocabSize, decode fixture, and changelog; report pass or fail on #1321 and update or close #912 as appropriate.

由索引模型根据 Issue 内容生成。

描述

assessment size:s skill:review sub-issue

Sub-issue of #1321 (HF tokenizer.json parity for BPE encoders).

Lane: 6 · Review
Skill needed: DARC review — must not have implemented any Lane 2 or Lane 4 task. Kotlin reading skills and access to a Python environment to re-run the goldens.
Size: s (a few hours)
Blocked by: all Lane 2 and Lane 4 sub-issues merged.

What to do

  1. D — Document. Re-read the parent's problem statement against the shipped code: are the four gaps closed, or did any lane narrow its scope? Record deviations as comments on the parent.
  2. A — Assess. Regenerate the goldens with the pinned tokenizers version (Lane 4 script) and run ./gradlew :skainet-io:skainet-io-core:downloadTokenizerJsonGoldenFixtures :skainet-io:skainet-io-core:jvmTest :skainet-io:skainet-io-core:linuxX64Test. Spot-check three cases by hand with the Python library.
  3. R — Research. Click every link in the parent, the docs page and the PR descriptions; confirm the tokenizers semantics cited (added-token matching, merges form) still hold at the pinned version.
  4. C — Code. Check that unsupported normalizer / pre-tokenizer types fail loudly, that vocabSize includes added tokens, that the #912 decode fixture is present, and that CHANGELOG.md lists the behaviour changes.
  5. Close #912 if its four asks are met, or comment on it with what remains.

Acceptance

  • Written review comment on #1321 with a pass/fail per acceptance criterion of the parent
  • Docs support matrix matches the code
  • #912 closed or updated
  • Result reported back on #1321 (then the parent can be closed)

Notes

@DarcValidated targets TensorOps functions and does not apply to the tokenizer package; the review comment on the parent is the durable record here.

主要语言
Kotlin
星标
52
派生
15
平均合并
1 天 15 小时
30 天内合并 PR
36

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。