INDEX

Explanations

. followed by pronouns or new clauses

The neuron flags imperative calls to action—phrases inviting the reader to do something (e.g. “check out,” “take a look,” “go and…”).

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

</h5>

-1.35

玊

-1.34

没有了

-1.27

それを

-1.25

where

-1.25

の場合

-1.22

Which

-1.20

 первые

-1.20

 วัน

-1.17

льта

-1.16

POSITIVE LOGITS

 voordat

1.39

and

1.37

 fué

1.27

又不

1.27

</strong>

1.26

 from

1.20

('../../

1.20

gemini

1.20

 veldig

1.20

却又

1.19

Activations Density 0.062%