INDEX

Explanations

numbers followed by time units

The neuron is specialized to detect numeric measurement tokens (e.g., quantities, times, temperatures) in instructional text.

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

 this

-1.82

 these

-1.77

sandalias

-1.70

medriver

-1.67

 such

-1.65

 equipada

-1.59

 zowel

-1.58

 ABBRE

-1.55



-1.52

摳

-1.50

POSITIVE LOGITS

 или

1.59

or

1.59

 یا

1.45

或

1.45

zości

1.44

你的

1.43

约翰

1.43

我們

1.43

或者

1.42

ก่อน

1.41

Activations Density 0.021%