INDEX

Explanations

Oops and apologies for minor mistakes

This neuron detects apology or error-interjection words (e.g. “Oops,” “Sorry,” etc.) at the start of a segment.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 äußerst

0.56

ellant

0.56

')$

0.55

toHave

0.55

archiviato

0.53

og

0.52

 ")");

0.52

 Mitgl

0.52

vole

0.52

 مؤرشف

0.52

POSITIVE LOGITS

sorry

0.67

 sorry

0.66

 rigorous

0.66

Sorry

0.63

पया

0.62

 radiography

0.59

 Sorry

0.58

 Generally

0.58

 restorative

0.57

 cores

0.57

Activations Density 0.001%