INDEX

Explanations

path, env, name, node, db, dir

The neuron fires on ALL-CAP sequences (often with underscores), i.e. constant-style identifiers or acronyms in 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

啄

-1.40

 with

-1.38

доступ

-1.30

%");

-1.29

 Seperti

-1.25

útil

-1.25

 Krzysz

-1.23

</h2>

-1.21

 версия

-1.17

entran

-1.15

POSITIVE LOGITS

儷

1.13

 STILL

1.13

 درباره

1.13



1.13

ALWAYS

1.13

 españa

1.09

 oeuvres

1.08

 alway

1.07

 souha

1.06

 Pyrénées

1.05

Activations Density 0.158%