INDEX

Explanations

unable to do

This neuron detects the infinitival “to” (especially in adjective-to verb constructions like “able to,” “unable to,” “difficult to,” etc.).

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

hlten

-1.17

 möglichst

-1.16

катерина

-1.13

 ingrédients

-1.09

⤒

-1.07

𝅥

-1.07

 predominantly

-1.04

 défauts

-1.03

Prefer

-1.02

 provas

-1.02

POSITIVE LOGITS

or

1.98

 even

1.84

 anymore

1.73

any

1.57

 ever

1.50

 access

1.42

 easily

1.37

 really

1.37

 anything

1.36

 because

1.34

Activations Density 0.065%