INDEX

Explanations

mercy and contexts of its use

The neuron fires on terms expressing mercy, leniency, or polite pleas (e.g. “mercy,” “leniency,” “please,” “patient,” “kind”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

伎

-0.75

rées

-0.74

щаются

-0.72

weli

-0.71

DECREF

-0.71

 healthful

-0.71

事了

-0.71

tonic

-0.70

glVertex

-0.69

loadData

-0.69

POSITIVE LOGITS

 mercy

3.80

Mercy

2.59

 Mercy

2.58

mercy

2.55

 merciful

2.13

 merciless

1.82

 MERC

1.72

 clemen

1.49

 compassion

1.48

 мило

1.45

Activations Density 0.011%