INDEX

Explanations

look at

The neuron fires on occurrences of the investigative phrase “look at” (including variants like “looking at,” “looked at,” and “looking into”).

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

When

-1.73

umenta

-1.71

 Neither

-1.59

These

-1.59

년

-1.58

-1.55

 Until

-1.55

 Except

-1.52

’)

-1.51

During

-1.50

POSITIVE LOGITS

how

2.13

</em>

1.71

笮

1.69

癭

1.56

⇹

1.56

 closely

1.54

 více

1.52

1.46

盉

1.46

 apresentam

1.44

Activations Density 0.040%