Repository navigation
Examples/TextClassify: the inbox app — a support inbox sorted on one tap, on iPhone and Mac - #65
Merged
Merged
Conversation
…tap, on iPhone and Mac App/ is the classifier as a screen: a synthetic support inbox, one tap on Sort inbox, and every message gets an intent, an urgency and a sentiment from one forward of gliner2.5-decide. The counts grow as the answers arrive; the time on screen is measured by the app (median ms per message, messages per second, the total). Sources/InboxCore holds the generator (an invented shop, plain item names, first names only, seeded) and InboxSorter, which the app and `textclassify-cli --inbox <count>` both run, so the screen and the CLI report the same numbers. `-autoplay 1 -log 1` runs it unattended and writes the result as JSON (with the per-message series), for a recording or a phone read back over devicectl. Measured 2026-09-26, seed 7: 1,000 messages in 40.9 s on an iPhone 18 Pro (iOS 27.0, GPU), 39.2 ms median per message, all on the 256-token graph; 30.1 s on an M4 Max (30.0 ms). The phone holds about 38 ms per message for the first 20 s of a run and slows from there. Examples/README.md gains the TextClassify row; CHANGELOG Unreleased gains the entry.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Examples/TextClassify/App/is the classifier as a screen: a support inbox of synthetic messages, one tap on Sort inbox, and every message gets an intent, an urgency and a sentiment from one forward ofgliner2.5-decide. The counts grow as the answers arrive, and the time on screen is measured by the app (the median milliseconds per message, messages per second, the total), on iPhone and Mac.What changed
Sources/InboxCore/:Inbox.swiftgenerates the inbox from a seed (an invented shop and app subscription, plain item names, first names only, no addresses or contact details);InboxSorter.swiftasks every message the three questions. The app and the CLI both run it, so the screen and the CLI report the same numbers.App/: the SwiftUI app (InboxScreen,InboxModel,Autoplay,TextClassifyApp),project.ymland the generatedTextClassify.xcodeproj.-autoplay 1 -log 1runs it unattended and writesDocuments/inbox-result-<epoch>.json(the total, the per-message series, the label counts, the thermal state before and after).CLI/main.swift:--inbox <count> [--seed N] [--dump] [--jsonl <path>]sorts the same inbox headless and prints one JSON line.Package.swift: theInboxCoretarget;Package.resolvedfor the example.README.md: the inbox app section.Examples/README.md: the TextClassify row.CHANGELOG.md: the Unreleased entry.Measured 2026-09-26, seed 7 (the README's numbers): 1,000 messages in 40.9 s on an iPhone 18 Pro (iOS 27.0, GPU), 39.2 ms median per message, all on the 256-token graph; 30.1 s on an M4 Max (30.0 ms).
Rebased onto main (d58aecf), no conflicts. CI's
example-app-releasebuilds Transcribe only, so this app was built locally: Release forgeneric/platform=macOSandgeneric/platform=iOS(CODE_SIGNING_ALLOWED=NO), andswift build -c releaseinExamples/TextClassify.