Commit Graph
19 Commits
Author SHA1 Message Date
ays953 815dcf979d Release v0.9.1 2026-11-13 12:06:24 +08:00
ays953 9cac3467a0 Release v0.9.0 2026-11-13 11:59:53 +08:00
ays953 fcc0f9d2db Release v0.8.0 2026-11-13 09:56:18 +08:00
ays953 6b4ec39dc1 Release v0.7.1 2026-11-11 17:01:37 +08:00
ays953 251a16fa08 Release v0.7.0 2026-11-11 16:29:06 +08:00
ays953 3670305f0f Release v0.6.3 2026-11-11 16:10:59 +08:00
ays953 1dacb8f164 Release v0.6.2 2026-11-11 12:13:15 +08:00
ays953 bb9cb0c4e0 Release v0.6.1 2026-11-11 12:00:10 +08:00
ays953andClaude Opus 5.5 2c5c70c066 Learn the previous-word context; fix first-tone characters after English
After an English word, 23 of 50 common first-tone characters (機, 吃, 書…)
were read as English when space was pressed: the fixed +1.5 context bonus
was far too strong for the trained weight scale, and the training data never
contained the "space just pressed on a first-tone syllable" decision.

Context is now the ctx.zh / ctx.lit features, trained with the rest; the
synthetic data includes the previous mode and the first-tone space decision.
Test accuracy: commit 98.22% -> 99.52%, typing 88.73% -> 91.49%, and when the
language differs from the previous word 92.0% -> 99.3%.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-11 11:59:57 +08:00
ays953 17f7d6f50d Release v0.6.0 2026-11-11 11:44:38 +08:00
ays953 fa0b78e5a4 Release v0.5.1 2026-11-11 11:36:12 +08:00
ays953 82f20a6551 Release v0.5.0 2026-11-11 10:37:37 +08:00
ays953 988ed471be Release v0.4.1 2026-11-11 10:18:14 +08:00
ays953andClaude Opus 5.5 6c36c01f0e Stop declaring Latin repertoire so password fields bypass the IME
With "Latn" in tsInputMethodCharacterRepertoireKey macOS treated AutoZhuyin
as ASCII-capable and kept it active in secure text fields, so typed passwords
showed up as underlined composition text. Apple's Zhuyin and vChewing declare
only Hant (not ASCII-capable); macOS then switches to an ASCII keyboard layout
in password fields and the input method never sees the keystrokes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-11 10:18:01 +08:00
ays953 42a2d3a2a3 Release v0.4.0 2026-11-11 09:57:34 +08:00
ays953 21142c2d98 Release v0.3.0 2026-11-11 08:58:43 +08:00
ays953andClaude Opus 5.5 476858d841 Learn classifier weights from synthetic typing data
Scoring is now features x weights (Features.swift / features.py), with the
old hand-tuned values as defaults. training/ generates synthetic mixed
Chinese/English buffers from the lexicon and word lists and trains the
weights with an averaged structured perceptron over the same segmentation
DP the input method runs. On 10,000 test buffers commit-time accuracy goes
from 97.58% to 99.62%; typing-time (prefix) accuracy from 86.15% to 87.81%.

Both apps load Resources/classifier-weights.json. A parity test checks the
Swift segmentation matches the Python one on 400 fixtures.

Also fixes space after a toned syllable being scored as an extra stray tone,
which could turn short Chinese like "u4 " into English.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-11 08:58:08 +08:00
ays953 0a7242ddb3 Release v0.2.0 2026-11-10 16:21:42 +08:00
ays953 582dddabaa first 2026-11-10 10:50:04 +08:00