INDEX

Explanations

positive testimonials with help

The neuron detects quoted testimonial or review passages—especially first-person praise statements enclosed in quotation marks.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 will

-1.86

is

-1.62

 hvil

-1.33

 might

-1.26

已經

-1.24

 with

-1.18

 volks

-1.17

已经

-1.15

 dager

-1.13

 ممکن

-1.12

POSITIVE LOGITS

chengladbach

1.72

 each

1.64

styleable

1.54

 semua

1.47

 helpen

1.41

 overall

1.41

每一次

1.40

Svara

1.39

 ['./

1.38

</i>

1.34

Activations Density 0.065%