INDEX

Explanations

not specifying or sharing

The neuron flags words expressing refusal or withholding of information (e.g. “refused,” “divulge,” “did not share”).

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

των

-0.92

reported

-0.91

ява

-0.88

ТОВ

-0.85

 pertinente

-0.83

VALID

-0.82

 skrev

-0.82

AppendLine

-0.81

understood

-0.81

⌐

-0.81

POSITIVE LOGITS

 specify

2.11

 specifies

1.91

具体

1.78

 specifics

1.77

 specified

1.71

 specifically

1.70

 specific

1.66

 provide

1.64

 specifying

1.63

say

1.60

Activations Density 0.028%