INDEX

Explanations

map and path

The neuron fires on tokens that include digits—typically technical identifiers, version numbers, or code variable names containing numbers.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 ﺍﻟ

1.47

 Rugby

1.31

 Veterans

1.16

TED

1.13

 Çok

1.10

 Lebih

1.09

 Labour

1.08

 VERY

1.08

ـ

1.07

POSITIVE LOGITS

and

1.16

with

1.04

to

1.00

initializer

1.00

configs

0.99

calculator

0.98

dataframe

0.98

for

0.97

name

0.96

regex

0.95

Activations Density 0.057%