INDEX

Explanations

dilemma and difficult choice

The main thing this neuron does is detect occurrences of the word “dilemma” (including its plural).

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

廰

-2.66

 probabilmente

-2.45

You

-2.42

は

-2.23

 duże

-2.22

後記

-2.20

 habitually

-2.19

我一直

-2.17

phòng

-2.14

 esclusivamente

-2.11

POSITIVE LOGITS

4.25

2.48

2.41

the

2.33

倬

2.23

2.22

羣

2.22

 годах

2.05

THE

2.05

2.03

Activations Density 0.004%