INDEX

Explanations

The neuron fires on second-person imperative phrases that prompt the reader to take action (e.g. “go ahead,” “try to”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 maupun

1.04

 sowie

0.97

 sekä

0.85

 oraz

0.77

 illetve

0.76

 valamint

0.73

 तसेच

0.72

 lẫn

0.72

 તેમજ

0.67

 అలాగే

0.66

POSITIVE LOGITS

并

1.23

並

1.02

并在

0.96

并

0.92

并且

0.88

 ולה

0.84

並

0.82

並且

0.80

And

0.78

并将

0.75

Activations Density 0.230%