INDEX

Explanations

diverse categories of study

The neuron specifically fires on the adjective “different,” particularly when used in scholarly phrases like “different experimental models,” “different areas,” “different methods,” 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

 Schluß

-1.55

 nebst

-1.52

所有人都

-1.52

 razu

-1.48

寥

-1.45

 alcune

-1.45

 were

-1.45

 quelli

-1.43

 Erfol

-1.42

unidade

-1.42

POSITIVE LOGITS

</em>

1.80

枓

1.72

ൎ

1.64

 måde

1.62

墻

1.59

閙

1.58

 arbeta

1.54

獓

1.54

 bague

1.52

 auttaa

1.51

Activations Density 0.047%