INDEX

Explanations

PEG, BST, ATR

The neuron is specialized in picking up tokens that are numeric measurements (especially decimal values) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

怍

-3.25

++){

-2.95

}+\

-2.81

has

-2.77

}$

-2.72

tolsó

-2.70

[]

-2.67

 apocalyptic

-2.66

ariously

-2.64

 $^{\

-2.61

POSITIVE LOGITS

“

3.84

on

3.08

2.98

2.63

疍

2.58

an

2.50

er

2.45

 frequentemente

2.41

 WHICH

2.27

 verminder

2.23

Activations Density 0.032%