INDEX

Explanations

adjectives or quantifiers followed by nouns

This neuron detects technical qualifier words—adjectives or modifiers (often in compound terms) that specify scale, degree, or type (e.g. “common,” “single,” “sub-,” “false,” “large,” “near,” etc.).

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

かけた

-1.63

 särsk

-1.52

における

-1.42

人が

-1.37

</em>

-1.36

 umgangssprach

-1.35

```

-1.34

 üblichen

-1.34

##

-1.33

 Rücksicht

-1.31

POSITIVE LOGITS

 natuurlijk

1.57

↵↵↵↵

1.54

not

1.52

↵↵↵

1.46

<bos>

1.38

 will

1.31

 confirmación

1.27

↵↵↵↵↵

1.23

 opção

1.23

锄

1.23

Activations Density 0.412%