INDEX

Explanations

reported speech and quotes

The neuron strongly activates on words that occur immediately before a sentence‐ending period (i.e. the last word of a sentence).

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

 güçlü

-0.94

suya

-0.88

 неправи

-0.88

 bahagia

-0.86

 décide

-0.86

vermectin

-0.85

—(

-0.85

Inhoud

-0.85

 corren

-0.85

─

-0.84

POSITIVE LOGITS

 ส่วน

1.06

 said

1.05

Similarly

1.03

 Similarly

1.03

 likewise

0.99

 Likewise

0.94

Likewise

0.91

 ditto

0.89

ὠ

0.89

MOUTH

0.88

Activations Density 0.016%