INDEX

Explanations

package names or module paths

This neuron fires most strongly on tokens that are part of Java (or similar) package/import qualifiers—i.e. company or organization names and domain-like identifiers 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

 (!__

-1.51

 like

-1.46

︲

-1.41

⎠

-1.34

嫦

-1.27

 également

-1.25

 ſuch

-1.25

）-

-1.24

 februar

-1.20

 essentiel

-1.20

POSITIVE LOGITS

</h3>

1.55

↵↵

1.34

 when

1.30

ά

1.25

sssss

1.20

 only

1.17

 masiva

1.17

six

1.16

</blockquote>

1.16

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.16

Activations Density 0.086%