INDEX

Explanations

The neuron detects formal framing and section-heading phrases typical of academic or patent texts (e.g. “the present invention,” “main contribution,” “field of”).

this invention or paper

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

 berbagai

-1.16

 from

-1.10

 ситуация

-1.09

 complicada

-1.07

 они

-1.06

 его

-1.05

 vielleicht

-1.03

אלה

-1.03

 izquier

-1.03

 einige

-1.03

POSITIVE LOGITS

 this

2.64

our

1.65

 herein

1.61

this

1.42

 here

1.38

今回

1.32

 данного

1.30

本文

1.23

นี้

1.23

 данной

1.20

Activations Density 0.170%