Repository navigation
Series temporales, clustering y mejoras UI - #2
Merged
Merged
Conversation
…ance training functionality - Changed button variant from "outline-info" to "outline-primary" in ModelReviewImageClassification component. - Enhanced training functionality in trainImageClassifier to support progress callbacks for epoch end and stopping conditions. - Updated I_MODEL_IMAGE_CLASSIFICATION and I_MODEL_IMAGE_28X28 to accept progress parameters during model training. - Added theme support for alerts in AlertHelper with a consistent primary color. - Introduced tests for N4LConfusionMatrix, DataFrameUtils, and training session functionalities. - Added dataset validation tests to ensure CSV files match expected structures and sources. - Improved page tests for Datasets and Glossary to cover modal interactions and variable displays.
- Implement tests for detecting and changing TensorFlow backends in `tfBackend.test.ts`. - Add tests for the `MODEL_CAR` and `MODEL_IRIS` classes to validate model outputs and class indexing in `MODEL_CAR.test.ts`. - Enhance `MODEL_LYMPHOGRAPHY` tests to verify class indexing and output consistency. - Create tests for `ModelReviewForms` to ensure proper rendering and data handling in `ModelReviewForms.test.tsx`. - Introduce diagnostics tests for graphics features and recommendations in `debugDiagnostics.test.ts`. - Update Vite configuration to resolve TensorFlow.js imports correctly and optimize build process.
…ara móvil Rendimiento - Los tres entrenamientos (clasificación tabular, regresión e imágenes) van en un worker: la página no se bloquea mientras se entrena. - El análisis de /analyze, la lectura de los ficheros subidos y las imágenes de MNIST/KMNIST también van en workers. - Los shaders se compilan antes de usarlos: la primera clasificación y la primera explicación con LRP ya no bloquean. - SHAP más rápido, gráficos y tablas que se pintan cuando el navegador está libre. - La build de producción ya no lleva React de desarrollo (quitado NODE_ENV de los .env). Datos - /analyze es ahora un AED completo: resumen, variables, distribuciones, relaciones, tabla con consultas y preprocesado con descarga del CSV. Se puede enlazar con ?dataset=<clave>. - Al subir un fichero se aceptan CSV con "," o ";" (con coma decimal), ARFF, JSON, JSONL y Parquet. - /datasets: enlace de cada conjunto al AED, tablas que se ven bien en el móvil y pestañas de datos y estadísticas en "Ver información". - Tablas de Plotly con los colores del tema y la columna objetivo resaltada. Interfaz - Barra de navegación fija, que marca la página actual y se cierra al navegar en el móvil. - Migas de pan en una línea, con salto entre el modelo preentrenado y entrenar con su conjunto de datos. - Índice de secciones como barra arriba en el móvil y cabeceras de tarjeta que no se salen de la pantalla. - Glosario nuevo: índice, buscador, una tarjeta por término y gráficas animadas de las funciones de activación; corregidas algunas fórmulas y añadidas las métricas. - Se puede dibujar con el dedo en el lienzo de MNIST y el visor no se abre al entrenar en el móvil. - Mejores formularios y selectores de ejemplos en las páginas de los modelos. - Arreglado el error al cambiar de modelo desde las migas de pan (Iris → Car). - Textos nuevos en español, inglés y japonés. Pruebas y despliegue - Pruebas en el navegador con Playwright (pnpm test:e2e) y más tests unitarios. - Nuevo workflow de comprobaciones (lint, tipos, tests y Playwright) que se ejecuta en cada pull request y antes de publicar en Netlify. - Nuevas dependencias: hyparquet, hyparquet-compressors, papaparse, @playwright/test y @types/papaparse. - Eliminados los componentes antiguos de DataFrame.
… improve comments
…ssion model review, improve object detection webcam handling - Updated `collectDiagnostics.ts` to export WebGL and system probing functions with additional documentation. - Modified `ModelReviewRegression.tsx` to remove unused DataFrame plot and add an analysis tab for exploratory data analysis. - Enhanced `ModelReviewRegressionDataset.tsx` to include an analysis tab with essential insights from the dataset. - Improved `ModelReviewObjectDetection.tsx` to handle webcam maximization and added a new `WebcamDeviceInfo` component for detailed device stats. - Created `WebcamDeviceInfo.tsx` to display real-time information about the webcam and system diagnostics. - Added tests for `N4LMaximizeButton` and `WebcamDeviceInfo` to ensure proper functionality and state management. - Updated `vite.config.ts` to optimize vendor chunking for better performance.
- Introduced LearningSettings component to manage step-by-step learning feature. - Updated Settings page to include LearningSettings alongside other settings. - Refactored tutorial reset functionality to clear guide progress and reset training tutorials. - Added tests for step-by-step settings and ensured proper functionality in Settings page tests. - Implemented new tests for N4LDatasetProcessForm, N4LDatasetViews, and N4LStepByStep components. - Enhanced model review and trainer guides to conditionally display step-by-step options based on settings.
…er experience - Enhanced `resetAllSettings` function to also clear trained models, download consent, and learning path data. - Added new tests for download consent and learning path functionalities to ensure proper behavior. - Introduced new test cases for training diagnostics and evaluation metrics to improve training feedback. - Implemented comprehensive tests for scroll actions and layer checks to ensure navigation and model integrity. - Updated trainer guide and layers guide tests to validate language translations and guide content usage.
- Changed import statements from double quotes to single quotes in various files for consistency. - Updated type definitions in `types.ts` to replace `Layer_t` with `ImageLayer_t` for better clarity in image processing layers. - Added new tests for KMeans clustering and training report functionalities. - Enhanced existing tests for datasets and pages to ensure proper functionality and coverage. - Adjusted Vite configuration for improved module resolution and consistency in code style.
- Implemented a new Task page to navigate between pretrained models and dataset design. - Enhanced the Model Review for Image Classification to display character forms with modern and old representations. - Created a new component for displaying character forms, including their readings and origins. - Updated the KMNIST model to include character forms and their respective images. - Added tests for the new Task page and character forms functionality. - Updated analytics to handle new task routes.
- Updated dataset download tests to include N4L variables and ensure proper file coverage. - Modified MenuSelect tests to reference model names from N4L packages. - Adjusted ModelReviewForms tests to utilize N4L model instances and ensure correct label translations. - Enhanced Pages tests to validate dataset links and ensure correct references to N4L datasets. - Refactored analyzeProjectDatasets tests to accommodate N4L dataset paths and ensure unique dataset keys. - Updated layersGuide tests to include new tests for IMAGE-CIFAR10 and validate layer configurations. - Revised modelReviewGuide tests to incorporate N4L datasets and ensure proper CSV feature extraction. - Added Vite configuration to support dynamic loading of N4L packages. - Introduced n4lPackages plugin to read and validate N4L packages from the public directory.
…eries features - Implement tests for detection summary in `detectionSummary.test.ts` - Create tests for face mesh rendering and person side detection in `faceMesh.test.ts` - Add glossary target tests to ensure correct mappings in `glossaryTarget.test.ts` - Introduce hand sign detection tests in `handSign.test.ts` to validate sign language recognition - Enhance layers guide tests for model weight calculations in `layersGuide.test.ts` - Extend model review guide tests to include time series models in `modelReviewGuide.test.ts` - Add smoothing tests to ensure stable detection outputs in `smoothing.test.ts` - Create time series guide tests to validate training steps and translations in `timeSeriesGuide.test.ts` - Update test setup to mock scrollIntoView for jsdom compatibility in `setupTests.ts`
- Deleted unused model files for breast cancer and salary regression. - Removed MNIST model files and their associated weights. - Added functionality to clear saved models selectively based on the model type. - Enhanced UI for displaying saved models with delete options. - Updated settings page to manage saved models more effectively. - Improved tests for model deletion to ensure correct functionality. - Configured Vite for HTTPS support to enable camera access from local devices.
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
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.
Summary
This PR introduces a broad set of improvements to Nets4Learning, covering new features, model management, training workflows, analytics, accessibility, and testing.
Main changes
Playground and UI: Refactored playground components, improved analytics tracking, added an animated 404 page, and enhanced the 2D lab, model code viewer, training predictions, and sortable layers.
Machine learning: Added new AED functionality, additional dataset formats, KMNIST examples, step-by-step learning options, and improved MNIST/CIFAR-10 training timeout configuration.
Model management: Integrated N4L packages, removed deprecated models, and enhanced model management capabilities.
Diagnostics and usability: Improved WebGL and system diagnostics, regression analysis, configuration settings, documentation, glossary, and accessibility.
Testing and reliability: Added unit tests for TensorFlow backend management, model predictions, object detection, face mesh, hand signs, and time series. Added an MNIST end-to-end training test and improved timezone handling and test timeouts.
CI/CD: Updated GitHub Actions workflows to improve deployment and environment configuration.
Code quality: Refactored code style and updated types across the project.
Testing
Expanded unit, end-to-end, and accessibility test coverage to improve reliability across the application.