INDEX

Explanations

strength and of

The main thing this neuron does is detect occurrences of the word “strength.”

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

</strong>

-2.61

铙

-2.38

喾

-2.36

栳

-2.36

 あります

-2.17

…

-2.08

aciones

-2.08

 ͡°

-2.05

 invented

-2.02

 forwarded

-1.98

POSITIVE LOGITS

of

2.92

⚢

2.63

または

2.38

2.34

～。

2.34

metery

2.25

镚

2.23

];

2.22

〜。

2.22

 eller

2.17

Activations Density 0.009%