INDEX

Explanations

access credentials

The neuron activates on occurrences of the word “credentials” (and its sub‐tokens).

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

enablog

-3.52

盌

-3.20

꜆

-3.16

窳

-3.09

 évaluation

-2.81

翃

-2.80

}$

-2.66

 odeur

-2.58

 activité

-2.56

珹

-2.55

POSITIVE LOGITS

”،

2.61

)”

2.47

但是在

2.34

Surprisingly

2.33

是如何

2.33

但却

2.30

’

2.28

酝

2.25

”;

2.19

却又

2.19

Activations Density 0.003%