INDEX

Explanations

somethingquestion

The neuron reliably activates on the vague filler word “something.”

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

Ингредиенты

-1.33

只是

-1.30

天

-1.30

จาก

-1.30

 semplicemente

-1.29

</h2>

-1.24

-1.23

gegeven

-1.23

 otras

-1.22

 broder

-1.21

POSITIVE LOGITS

you

1.97

1.95

 than

1.76



1.55

 takimi

1.46

 that

1.43

 sprawd

1.42

等我

1.42

even

1.40

僕も

1.38

Activations Density 0.027%