qlaya_flutter 0.5.4

QLaya Flutter & Dart client for the fast on-device decision engine powered by the qlaya.int8.onnx model for classification, scoring, verification, and routing tasks.

0.5.4

• Static Analysis & Lint Fixes: Resolved all pana static analysis warnings including dangling library doc comments and unescaped angle brackets in CLI doc comments.
• Code Formatting & Lint Rules: Added analysis_options.yaml adhering to package:lints/recommended.yaml and formatted all codebase files with dart format.
• Clean Dartdoc: Ensured 100% warning-free and error-free documentation build.

0.5.3

• pub.dev Image Compatibility: Updated cover banner in README.md to use direct GitHub CDN asset URL for cover.png, ensuring reliable rendering on pub.dev and GitHub.

0.5.2

• Visual Cover Banner: Added official vector-wrapped cover.svg hero banner to package documentation.
• Streamlined Repository: Added model file patterns to .gitignore and removed model artifacts from version control.
• Documentation Polish: Refined README structure and visual presentation.

0.5.1

• Interactive Flutter Example App: Replaced CLI script in example/main.dart with a complete Material 3 Flutter application featuring interactive playgrounds for Classification, Urgency Scoring, Verification, and Workflow Routing.
• Model Download UI: Added live progress tracking for Hugging Face model downloads in the Flutter example.
• Simplified Documentation: Overhauled README.md to guide users step-by-step from installation to basic usage.
• Cleaned API Interfaces: Removed legacy server configuration requirements from docs and client descriptions.

0.5.0

• Hugging Face Hub hosting: Model weights are hosted on Hugging Face at saipy10/qlaya with direct download support.
• CLI Downloader: Added dart run qlaya_flutter:download to download qlaya.int8.onnx directly from terminal with progress indicators and speed display.
• Programmatic Downloader: Added QLayaDownloader to verify, download, and ensure model availability directly within Flutter or Dart code.
• Optimized package size: Package published to pub.dev is lightweight (<20 KB) while weights are fetched on demand.
• New Task API: Added QLayaTasks and four typed task classes for common decision patterns:
  • QLayaClassificationTask / client.classify() — intent and category classification.
  • QLayaScoreTask / client.score() — continuous scoring and rating.
  • QLayaBoolTask / client.verify() — boolean verification.
  • QLayaRoutingTask / client.route() — workflow and destination routing.
• client.runTask(task): Generic typed task runner for any QLayaTask<T>.
• QLayaModels.int8 / QLayaModels.defaultModel: Model specification with Hugging Face repository and direct download link.
• New result types: QLayaClassificationResult, QLayaScoreResult, QLayaBoolResult, QLayaRoutingResult.
• Updated examples, tests, and documentation.

0.4.1

• Update package metadata and repository links to dedicated repository: saipy10/Qlaya-Flutter.
• Add badges, repository documentation, and issue tracker links.

0.4.0

• Initial release of qlaya_flutter for Flutter and Dart.
• Complete model registry (QLayaModels) with 10 user-selectable quantized model variants (INT8, INT4, FP16, student distillations) hosted on Hugging Face at saipy10/qlaya.
• QLayaClient HTTP client for connecting to QLaya inference servers.
• Type definitions for QLayaQuestion, QLayaAnswer, QLayaPrediction, and QLayaModelSpec.
• Includes runnable example demonstrating model variant resolution.