INDEX

Explanations

pronoun followed by verb

The neuron is keyed to personal address, activating strongly on first‐ and second‐person pronouns (“I,” “you,” 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

もあって

-1.71

 will

-1.70

年には

-1.51

もある

-1.42

 nacht

-1.36

 команде

-1.30

では

-1.28

による

-1.27

が

-1.25

も

-1.23

POSITIVE LOGITS

 From

1.53

 from

1.38

 this

1.23

And

1.12

ようで

1.11

to

1.10

 With

1.10

 Compared

1.07

 různých

1.06

を参照

1.03

Activations Density 0.046%