INDEX

Explanations

describing difficulty of actions

The neuron consistently activates on the token "to," effectively detecting the word “to” (often marking infinitives or prepositions).

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

-2.09

–

-1.89

But

-1.86

-1.82

 entwick

-1.73

Yet

-1.73

 ยัง

-1.73

不錯

-1.69

 were

-1.65

 deline

-1.64

POSITIVE LOGITS

鮡

2.17

놋

1.88

ﺅ

1.77

藹

1.75

-</

1.74

옻

1.74



1.74

caped

1.73

 використовувати

1.73

……”

1.69

Activations Density 0.027%