INDEX

Explanations

result or results

This neuron fires on mentions of reported outcomes, specifically occurrences of “result” or “results” (often introducing numerical findings).

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

Як

1.22

1.21

ab

1.19

առ

1.14

1.12

 morals

1.06

 Tactics

1.06

С

1.06

funk

1.05

POSITIVE LOGITS

 результаты

1.33

结果

1.30

 результат

1.29

 RESULT

1.26

result

1.25

 resulta

1.24

 result

1.19

 резулта

1.18

 결과를

1.16

 obtenus

1.14

Activations Density 0.067%