INDEX

Explanations

companion, partner, friend

This neuron picks up marketing-style promotional phrases (e.g. product “partners,” “friends,” “addictive” features, “personal” appeals).

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

的情况

0.71

 unmarked

0.69

 கிடைக்க

0.66

 thumbnails

0.66

 retaining

0.64

 acreage

0.62

 structures

0.62

 categories

0.61

的情況

0.61

 tags

0.61

POSITIVE LOGITS

 asistente

0.95

 ваш

0.88

 chatbot

0.87

chatbot

0.87

 aiuto

0.87

 możesz

0.86

 môžu

0.82

あなたの

0.82

 możemy

0.80

 energética

0.80

Activations Density 0.001%