Pipeline
Annotate
LLMLingua-2
Gemini Compact
LLMLingua-2
not run
Gemini Compact
not run
Evaluate
Parameters
Evaluating: —
Weights
Semantic fidelity
Logical coherence
Structural integrity
Readability
Reference text optional
Paste a gold-standard compression to score semantic fidelity against a specific target instead of the original input.
Custom dimensions scaffold
Custom scores optional
Each row becomes an additional radar axis. Updates live, no re-run needed.
Results
Run evaluate to see results.