INDEX

Explanations

end of descriptive phrases

The neuron flags the final content word just before the closing punctuation at the end of each short text segment.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 ज़िंदगी

0.41

ಿದ್ದೇನೆ

0.40

னெ

0.40

𝘃

0.39

)]:

0.39

 iniziare

0.39

nCurr

0.39

ْد

0.39

 اپنی

0.39

どんど

0.39

POSITIVE LOGITS

0.64

此外

0.59

。

0.58

0.55

."

0.54

.(

0.54

0.53

 Moreover

0.53

 Additionally

0.52

.¹

0.50

Activations Density 0.000%