Add/Edit a Score

Learn how to create and manage individual scores within scorecards using the Plexus dashboard interface.

Adding Scores in the Dashboard

Scores are individual evaluation criteria within a scorecard. The dashboard provides an intuitive interface for creating and configuring scores.

Step-by-Step Guide

  1. Access Score Creation:

    Open your scorecard and click "Add Score" or edit an existing scorecard.

  2. Choose Score Type:

    Select from available score types:

    • Sentiment Analysis
    • Content Quality
    • Grammar Check
    • Custom Metrics
  3. Configure Parameters:

    Set up the score configuration:

    • Score name and description
    • Weight (importance in overall scorecard)
    • Threshold (minimum acceptable score)
    • Custom parameters specific to the score type
  4. Preview and Test:

    Use the preview feature to test the score against sample content.

  5. Save Score:

    Click "Add Score" to include it in your scorecard.

Editing Existing Scores

  1. Locate the Score:

    Find the score you want to modify within your scorecard.

  2. Access Edit Mode:

    Click the edit icon next to the score.

  3. Modify Settings:

    Update the score's configuration as needed.

  4. Save Changes:

    Click "Save" to apply your modifications.

Score Version Management

Scores in Plexus support versioning, allowing you to track changes and manage different implementations:

Creating New Versions

When you edit a score and save changes, a new version is automatically created. You can add notes to document the changes made in each version.

Champion Versions

Each score has a designated "champion" version that is used for evaluations. You can promote any version to champion status when you're satisfied with its performance.

Featured Versions

Mark important versions as "featured" to highlight them in the version history. This helps track significant milestones in your score's development.

Score Configuration Tips

Weight Balancing

Carefully consider the relative importance of each score when setting weights. The total of all weights in a scorecard should equal 1.0.

Threshold Setting

Set appropriate thresholds based on your quality requirements and test with representative content samples.

Score Types

Choose score types that align with your evaluation goals. Combine different types to create comprehensive assessments.

Using the CLI

For automated score management, you can use the Plexus CLI:

# View detailed information about a score
plexus scorecards score "Score Name" --account "account-name"
plexus scorecards score "score-key" --account "account-name"

# Show version history and configuration
plexus scorecards score "Score Name" --account "account-name" --show-versions --show-config

# List all scores for a specific scorecard
plexus scorecards list-scores --scorecard-id "scorecard-id"

# Coming soon:
# View version history for a score
plexus scorecards history --account-key "account-key" --score-key "score-key"

# Promote a version to champion
plexus scorecards promote --account-key "account-key" --score-id "score-id" --version-id "version-id"

# Add a new score to a scorecard
plexus scores add --scorecard-id "card-id" --name "Quality Score" --type quality --weight 0.5

# List all scores in a scorecard
plexus scores list --scorecard "Quality Assurance"

# View score configuration
plexus scores info --score "Grammar Check"

Efficient Score Lookup

The score command supports multiple lookup methods:

  • By ID: plexus scorecards score "score-id"
  • By key: plexus scorecards score "score-key"
  • By name: plexus scorecards score "Score Name"
  • By external ID: plexus scorecards score "external-id"

You can scope the search to a specific account or scorecard for faster results.

Python SDK Reference

For programmatic score management, you can use the Python SDK:

from plexus import Plexus

plexus = Plexus(api_key="your-api-key")

# Get a scorecard using any identifier (name, key, ID, or external ID)
scorecard = plexus.scorecards.get("Quality Assurance")

# Get a score using any identifier
score = plexus.scores.get("Grammar Check")

# Get all scores in a scorecard
scores = scorecard.get_scores()

# Get score configuration
config = score.get_configuration()

# Get score evaluation results
results = score.get_results(limit=10)

Like the CLI, the Python SDK also supports the flexible identifier system, allowing you to reference resources using different types of identifiers.

YAML Configuration

Scores can be configured using YAML for advanced customization:

name: Quality Score
key: quality-score
externalId: score_123
type: LangGraphScore
parameters:
  check_grammar: true
  check_style: true
  min_word_count: 100
threshold: 0.8
weight: 0.5

Coming soon: The ability to pull and push YAML configurations using the CLI for offline editing and version control.

Coming Soon

Additional score features are being developed. Check back soon for:

  • New score types and metrics
  • Advanced scoring algorithms
  • Custom evaluation parameters
  • Score performance analytics
  • Bulk score operations
  • YAML synchronization for offline editing