尚未检测到更新。GitHub 发布版本后会显示在这里。
按最近已完成的 8 小时观测回合间的 GitHub 星标增量排序
Automate with computer-use agent (ComputerUse / CUA) for macOS, Windows, Linux, Browser, iOS & Android. It sees the screen, clicks, types, browses and writes the code too. One agent, every computer.
翻译准备中
Offline word data packs for DocGraph Languages (CC BY-SA 4.0, from Wiktionary)
翻译准备中
正在加载完整 README。
查看完整 README ↗Offline word data for DocGraph Languages (ADR 0163): one file per language, downloaded by the app when the user adds that language. German is the only pack today.
| Pack | Contents | Download |
|---|---|---|
dict:de-en (de-pack-v<N>.sqlite3.gz) | 93k nouns (gender, every case form), 14.9k verbs (conjugation, Perfekt auxiliary, separable prefix), 18k adjectives (comparison, declined forms), 99k German–English senses with gender and plural per noun homograph, a 567k-row form → lemma index | ≈ 26 MB (≈ 71 MB unpacked) |
Packs are published as GitHub Release assets (tag de-pack-v<N>), never
committed. Each release carries the pack and its manifest (size, sha256,
counts, sources). The app pins tag + sha256.
python3 -I de/build.py --version 1
python3 -I de/check.py build/de-pack-v1.sqlite3
Standard library only. The build downloads the sources pinned in
de/sources.json into sources/ and refuses any whose sha256 has changed.
check.py tests the pack against hand-verified grammar facts (verbs, nouns,
adjectives, noun homographs, inflected forms, senses), the homograph link
rate, sense de-duplication and the coverage of de/check-text.txt.
meta(key, value) — pack, version, schema, built, licence, credits, sources.lemmas(id, lemma, pos) — every noun, verb and adjective once
(pos = noun / verb / adj).nouns(id, genders) — m / f / n or several (m,n).verbs(id, aux, prefix, weak, pres_ich, pres_du, pres_er, past_ich, subj2_ich, imp_sg, imp_pl, part2, present, past) — aux is haben,
sein or both with the usual one first; present/past are JSON lists of
six forms (ich … sie) when Wiktionary has the full table.adjectives(id, comparative, superlative) — gut: besser, am besten.forms(form, id, tags, particle) — single-word form → lemma; tags such
as nom_pl,dat_pl, pres_er, or for adjectives pos / comp / sup;
particle holds a separated prefix (ruft + an → anrufen).senses(lemma, pos, hom, num, definition, translations, importance) —
WikDict senses, one row per distinct translation set of a homograph
(merged German definitions joined with |); hom separates homographs.homographs(lemma, hom, genders, plural) — gender and plural of each noun
homograph (See 1 = m lake, 2 = f sea; Bank 1 Bänke, 2 Banken),
matched to German Wiktionary by definition text, else the noun data's
gender when unambiguous (96 % of noun homographs).LICENSE).de/check-text.txt: CC BY-SA 4.0 (LICENSE-DATA.txt),
credits in NOTICE.md.Anyone redistributing or showing pack data (including DocGraph) must:
credit the sources as in NOTICE.md; name the licence with a link; say the
data was changed; keep adapted data under CC BY-SA 4.0; add no further
terms; and not lock the data behind DRM or encryption (a checksum is fine).