INDEX

Explanations

went to or on

The neuron specifically activates on occurrences of the past‐tense verb “went.”

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

er

-2.77

-2.38

 that

-2.30

 れる

-2.19

 increase

-2.02

 दौरान

-1.95

 instantaneously

-1.94

’;

-1.86

laszt

-1.86

’.

-1.86

POSITIVE LOGITS

 volna

2.34

2.22

淠

2.13

～！

2.03

to

1.98

佉

1.95

鈁

1.90

 administratif

1.87

 brill

1.84

 That

1.82

Activations Density 0.009%