INDEX

Explanations

I would + action

The neuron fires strongly on first‐person conditional recommendation phrases—especially “I would …” constructions.

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

鯔

-1.09

 когда

-1.07

时的

-1.00

 alerg

-0.96

杙

-0.94

ודים

-0.94

 décid

-0.93

 ibland

-0.93

our

-0.93

 Alembic

-0.93

POSITIVE LOGITS

 myself

1.53

 would

1.48

 then

1.14

 start

1.12

Would

1.06

 approach

1.05

try

1.05

be

1.05

 rather

1.04

 first

1.02

Activations Density 0.020%