dotnet / dotnet/machinelearning-modelbuilder
Suggestion: Add Optional "Advanced Settings" Section in Model Builder
- Vorherrschende Sprache
- Dockerfile
- Sterne
- 285
- Forks
- 66
- PR-Merge-Kennzahlen
- Keine gemergten PRs in 30 T.
Beschreibung
Model Builder is a great tool for building machine learning models without writing code. However, I think its potential is underutilized by not offering a section for more advanced users who want greater control over the training process.
**Proposal:**
Add an optional and hidden-by-default section called "Advanced Settings" or "Expert Mode", which includes the ability to:
- Manually choose the training algorithm.
- Edit key hyperparameters (e.g., number of trees, learning rate, number of epochs, etc.).
- Customize the data preparation pipeline (e.g., normalization, encoding, cleaning).
- Set limits for time or resources used during AutoML.
- Display more detailed evaluation metrics (e.g., confusion matrix, ROC curve, etc.).
**Why this matters:**
This would allow developers with more experience or technical curiosity to get more out of ML.NET without having to abandon Model Builder or rewrite the generated code.
**Benefits:**
- Keeps things simple for beginners.
- Unlocks more flexibility for intermediate and advanced users.
- Extends the usefulness and lifecycle of Model Builder within Visual Studio.
Thanks for the great work! I believe this feature could make Model Builder even more powerful and appealing to a broader audience.
Beitragsleitfaden
Für dieses Repository ist kein Beitragsleitfaden indexiert
Rechercherichtung
Das Issue nennt Model Builder, aber keine Dateien, Tests oder Einstiegspunkte. Beginne damit, die UI von Model Builder und den Ablauf der Trainingskonfiguration zu erfassen, und kläre anschließend, welche der vorgeschlagenen Steuerelemente im Umfang liegen und wie das standardmäßig ausgeblendete Verhalten sowie detaillierte Evaluierungsmetriken validiert werden.
Vom Indexierungsmodell aus dem Issue-Text verfasst.
Bewertung
- Tech-Stack
- machine-learning
- Bereich
- machine-learning
- Issue-Typ
- Feature
- Schwierigkeit
- 5/5
- Geschätzter Aufwand
- Über eine Woche
- Aktivitätsstatus
- Veraltet
- Klarheit
- Muss geklärt werden
- Anfängerfreundlichkeit
- 25/100