INDEX

Explanations

understanding the problem

The neuron detects discourse‐signaling cue words (e.g. “thing,” “problem,” “point,” “but,” “here’s,” “fact”) that introduce or emphasize transitions in an argument.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 अनुभवों

0.66

 esperienze

0.56

 cotid

0.55

 activités

0.55

 experiencias

0.54

の頃

0.54

頻

0.54

tutorials

0.54

 frequentemente

0.54

 paes

0.52

POSITIVE LOGITS

 problem

2.59

問題

2.47

问题

2.42

 problema

2.42

problem

2.27

 problème

2.25

 проблема

2.25

 문제는

2.22

 Problem

2.19

 PROBLEM

2.19

Activations Density 0.718%