LMM-Vibes Pipeline Results Explorer

🧵 StringSight Evaluation Console data

Load Data

Base Results Directory: data

Select Experiment

Choose an experiment to load its results

Select Models for Analysis

Choose which models to include in comparisons

all

Minimum Cluster Size

1 50

Only show clusters where the distinctiveness score is statistically significant

Show Only Frequency Significant Clusters

Only show clusters where the quality score is statistically significant

Show Only Quality Significant Clusters

Sort Clusters By

Top N Clusters per Model

1 10

Quality Metric

View Type

Model Quality Comparison

Switch view to Table or Plot as desired.

Select your experiment to begin.

·

Built with Gradio logo

·

·