INDEX

Explanations

gives a [noun]

The neuron is primarily firing on the single‐letter indefinite article “a” (and similarly on very short function words like “and”).

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

-2.09

-2.05

-2.03

-2.00

ي

-1.89

-1.85

<tr>

-1.74

-1.70

 könnten

-1.67

we

-1.66

POSITIVE LOGITS

and

2.41

 langsung

1.66

 bereiken

1.65

 sélectionner

1.63

 помогает

1.63

 "/";

1.61

to

1.60

mousse

1.60



1.57

atissimo

1.55

Activations Density 0.023%