Repository navigation
C 批：评测尺子 + 一级简码入表 + 有界位次上移（1.1.0） - #223
Merged
Merged
Conversation
Subjective complaints about candidate order cannot be verified, so this adds a ruler before touching ranking: a reproducible sample set derived from already-registered data (intent from jieba frequencies, codes from the WindInput-derived table, so the two are not the same source), an evaluator that drives the real Session, a make target, and the frozen baseline bound to commit plus table hashes. The baseline shows ranking within a code is healthy and the real gap is the first-level codes. Signed-off-by: wangzhizhou <824219521@qq.com>
Wubi users type one letter for the twenty-five most common characters, but the table had no codes shorter than three letters, so that key fell back to a prefix list where the intended character was usually neither first nor even present. The level-one table is now generated into the code table from a registered project-compiled list, which makes the key an exact hit without touching the ranking code. Personalization no longer adds a bonus to the static weight; it moves a learned candidate up one or two places, never past the current first choice and never across the exact/completion/fuzzy blocks. The evaluation set shows the first-level hit rate going from 56 to 100 percent with no other category changing. Signed-off-by: wangzhizhou <824219521@qq.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
C 批的两步：先造尺子（B 基线），再按尺子改。
docs/MVP后执行计划.md§3。1. 评测尺子（报告 227）
crates/ime-cli/src/bin/ime-eval.rs：真实Session+ 真实码表评候选位次，4 条单测tools/eval/gen-samples.py→samples.tsv（200 条，意图来源与排序权重不同源）make eval-candidates、tools/eval/baselines/1.0.2-build37.json（绑定提交/样本/码表 SHA）2. 一级简码入表（P0，数据）
data/sources/wubi86-level1-codes.tsv（D17，自建清单，CC-BY-SA-4.0，25 条）generate_core.rs追加 25 行，tools/check-level1-codes.py校验 25/唯一/全码一致3. 有界位次上移（P1，核心）
bonus = learned.min(32)与"单字 bonus = 0"评测结果
验证
make verifyPASS；离线 56 文件 / 303 例 PASS；新增 1 个简码用例 + 重写 3 个钉旧语义的用例。未做
二级 625 条需先登记 D17-bis；
jianma2现状不变（93.3% 首选），评测里单列便于将来对比。发布前仍需：装包自用 3 天 + IT-25 + IT-22 受影响行 + 升级闭环（一次换包做四项）→ 全绿才 tag
v1.1.0。