INDEX

Explanations

urls

The neuron fires on standalone number tokens (e.g. sequences of digits such as “450”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

-2.45

なども

-2.41

 zarówno

-2.38

 dando

-2.30

 secluded

-2.27

 habilitar

-2.16

等は

-2.16

 famed

-2.13

 denominado

-2.09

同

-2.06

POSITIVE LOGITS

啻

2.09

Transk

2.03

 bella

2.03

蕯

2.02

 their

1.97

1.91

 EVERY

1.90

ſtra

1.84

 годом

1.83

 them

1.81

Activations Density 0.027%