INDEX

Explanations

search URLs

The neuron never activates on any tokens, so it does not detect or respond to any particular pattern.

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

msk

0.66

Som

0.60

Italy

0.60

Portugal

0.59

Iraq

0.59

Thailand

0.59

Совет

0.58

kada

0.57

Poland

0.57

fought

0.57

POSITIVE LOGITS

 ejecutar

0.82

 executar

0.74

 ejec

0.66

 execute

0.64

 servlet

0.64

 buscar

0.63

 ejecuta

0.62

 busca

0.59

 lookup

0.57

 exec

0.57

Activations Density 0.001%