INDEX

Explanations

location-related words across languages

The neuron strongly activates on non-Latin (especially Indic) script tokens, effectively detecting words in languages like Hindi and Oriya.

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

 उसमे

0.81

 इसमें

0.79

 उसमें

0.79

 તેમાં

0.72

 여기서

0.70

 ఇందులో

0.70

 তাতে

0.70

ในสนาม

0.67

 समोर

0.65

 ใน

0.65

POSITIVE LOGITS

 bulunan

1.34

 रहकर

1.23

 befind

1.21

 অবস্থিত

1.15

 található

1.11

 থাকা

1.09

 ያሉ

1.04

 olev

1.01

 terdapat

0.99

 رہتے

0.97

Activations Density 0.111%