INDEX

Explanations

matching

The neuron detects occurrences of the word “match” (or its inflected form “matches”).

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

</em>

-2.33

$\

-1.88

++];

-1.88

)$,

-1.88

)$

-1.78

ньому

-1.73

AFFIRMED

-1.71

ubscribe

-1.68

):

-1.67

());

-1.66

POSITIVE LOGITS

the

2.72

 your

2.55

 législation

1.88

our

1.86

 réglementation

1.77

おすすめの

1.68

 behulp

1.65

了啊

1.58

 principali

1.56

 eficiente

1.56

Activations Density 0.021%