INDEX

Explanations

renamed, merged, became

This neuron detects subword pieces of formal named entities—especially multi-token segments of organization, place, or institution names.

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

 estão

1.05

 ہیں

1.04

 tiene

1.02

 můžete

1.01

 tienes

1.00

 tycker

0.97

 están

0.96

 chce

0.96

 môžete

0.93

 हैं

0.92

POSITIVE LOGITS

became

1.35

 became

1.34

 renamed

1.30

 merged

1.25

 reorganized

1.22

 formerly

1.19

 originally

1.10

 damals

1.10

 Became

1.10

 successor

1.09

Activations Density 0.704%