INDEX

Explanations

closing bracket `]`

The neuron fires on numeric literals used as indices inside square‐bracket array or pointer accesses.

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

-2.69

ness

-2.08

re

-2.06

-1.94

不断的

-1.92

 حتما

-1.89

ほうが

-1.87

 Mijn

-1.84

-1.82

POSITIVE LOGITS

꯳

2.13

 který

1.98

一流

1.91

sza

1.90

ichtete

1.84

dicionado

1.80

पुर

1.80

obedience

1.78

layah

1.78

}$,

1.78

Activations Density 0.012%