INDEX

Explanations

questions, reactions, advantage

This neuron responds to unusually long tokens—i.e. words of substantial character length.

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

resultados

0.47

muje

0.44

 excelencia

0.44

ahul

0.43

вичайно

0.43

 trazendo

0.43

が出来

0.42

 excellence

0.42

 epitome

0.41

 உதவுக

0.41

POSITIVE LOGITS

 with

0.66

on

0.60

to

0.59

 from

0.52

 với

0.51

for

0.50

 على

0.50

 tới

0.49

 onto

0.47

with

0.47

Activations Density 0.000%