INDEX

Explanations

hope this helps

The neuron fires strongly on second-person help/offering phrases—i.e. when the text directly addresses “you” with “help,” “helped,” etc., offering assistance or guidance.

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

 alertas

0.63

ையான

0.63

 Norwegen

0.61

 сла

0.61

 longitudinale

0.61

 Augmented

0.60

 precedenti

0.59

 FLORIDA

0.58

 habitaciones

0.57

 хоче

0.56

POSITIVE LOGITS

Brid

0.63

径

0.63

In

0.61

成员

0.58

Mission

0.58

Kxe

0.56

Directory

0.55

 simplify

0.55

Bridge

0.54

ෂ්

0.54

Activations Density 0.001%