INDEX

Explanations

HTML title tags

This neuron fires on non‐word tokens—that is, punctuation marks, numeric literals, file‐path separators, and other symbols rather than ordinary alphabetic words.

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

 Brink

-0.77

dfs

-0.73

mund

-0.73

tih

-0.71

duc

-0.69

 escribiendo

-0.66

اور

-0.66

quet

-0.66

 scoprire

-0.66

 incrível

-0.65

POSITIVE LOGITS

 Adams

0.77

 ADAMS

0.68

Adams

0.66

ロマ

0.65

 Invite

0.64

 VIII

0.63

Estas

0.63

ूह

0.61

 cellspacing

0.61

 gallows

0.61

Activations Density 0.072%