INDEX

Explanations

assignments with equals

This neuron activates on standalone numeric literals (integer constants) in the code.

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

 will

-1.59

for

-1.24

 while

-1.17

ori

-1.16

 WALTZ

-1.08

on

-1.08

 have

-1.07

é

-1.07

byes

-1.07

もありますが

-1.05

POSITIVE LOGITS

 sobrevivir

1.38

 robar

1.16

 trám

1.14

Literatura

1.13

gâ

1.12

 lác

1.12

ZIONI

1.11

ikistan

1.10

Sumber

1.09

Kaynak

1.09

Activations Density 0.004%