chore: bump DasherCore to v0.2.14 (complete) — 150 alphabets trained - #32
Merged
Conversation
- 150 alphabets train on real Tatoeba sentences (was 48) — 56 new corpora imported from WorldAlphabets v0.1.1 (CC-BY 2.0 FR) - Training weight: 74MB -> 23MB (400KB cap) — 51MB off the APK - Alphabet index records training_available - Script-aware corpus retargeting + injection into undeclared WA alphabets (54 now declare their Tatoeba corpus) Signed-off-by: will wade <willwade@gmail.com>
Submodule at ba7a634a — the full v0.2.14 including the 56 Tatoeba corpora that were initially eaten by .gitignore (DasherCore #74). Signed-off-by: will wade <willwade@gmail.com>
Two bugs discovered on-device/emulator: 1. Settings crash: the LP_UNIFORM parameter description contained the multiplication sign as a single Latin-1 byte (0xD7) instead of UTF-8 (0xC3 97) — 'Uniform probability weight (×1000)' — a leftover mojibake 'fix' that traded wrong bytes for broken encoding. JNI's NewStringUTF validates Modified UTF-8 and aborts the process the moment nativeGetParameterInfo returns the string. Fixed in DasherCore's Parameters.cpp (also fixed upstream). 2. Alphabet picker showing one entry: the engine was queried (getAlphabetNames, locale-follow) before the canvas laid out, and set_screen_size was gated on canvas width > 0 — so Realize() (which builds the 474-entry name index) never ran before the first query. The picker got GetPermittedValues' single-item fallback (the current alphabet name). DasherEngine.create() now calls nativeSetScreenSize(800, 600) as a provisional size so the engine realizes immediately; the canvas corrects it when it lays out. Verified on emulator: Settings opens without crashing; alphabet picker lists 475 alphabets (Abc through Afrikaans and beyond). Signed-off-by: will wade <willwade@gmail.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
1 participant
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.
Submodule at
ba7a634a— the complete v0.2.14 including all 91 training corpora (the original tag was missing 56 files eaten by .gitignore, fixed in DasherCore #74).training_availableVerified on emulator: 91 corpora in
files/dasher_data/training/, canvas rendering with weighted nodes.DCO signed.