INDEX

Explanations

transformer

The neuron strongly detects technical mentions of “transformer” (and its variants, e.g. “transformers,” “isolated transformer,” “leakage transformer,” “current transformer”) in an electrical‐engineering context.

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

-1.79

 extremely

-1.72

-1.63

 discrepancies

-1.59

 incredibly

-1.55

'-

-1.55

or

-1.53

 reddit

-1.53

 shimmering

-1.52

POSITIVE LOGITS

 JUST

1.65

YMS

1.59

 longstanding

1.56

 redé

1.53

 Füßen

1.51

蒞

1.51

 caos

1.47

خته

1.46

數字

1.46

 futuras

1.45

Activations Density 0.021%