INDEX

Explanations

adenocarcinomas, cheddar, sedan, deer

The neuron fires strongly on concrete, countable noun mentions—especially specific object categories (e.g. vehicles, foods, weapons, disease names).

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

ské

-1.38

("");

-1.37

の準備

-1.36

 להם

-1.34

</h4>

-1.33

いは

-1.30

árbol

-1.28

 campeón

-1.27

地板

-1.27

elena

-1.22

POSITIVE LOGITS

鬈

1.65

If

1.57

Only

1.57

VERTISEMENT

1.43

When

1.38

During

1.38

发表于

1.35

 only

1.34

 همان

1.34

KommentareTeilen

1.33

Activations Density 0.341%