INDEX

Explanations

code blocks with specific syntax

The neuron detects words that convey effort or trouble, such as “effort,” “trouble,” and “nothing was too much trouble.”

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

 حم

0.60

 أرض

0.59

ifestyle

0.56

 وان

0.55

ਭ

0.55

 சிவப்பு

0.55

 آئے

0.55

acán

0.54

[\

0.53

arları

0.53

POSITIVE LOGITS

 effort

1.08

 efforts

0.94

 Effort

0.93

0.90

effort

0.86

{-

0.82

{.

0.82

{'

0.81

 노력

0.81

{}{

0.80

Activations Density 0.004%