INDEX

Explanations

R&D, R&B, Wi-Fi, Q&A

The neuron fires on short uppercase letter sequences that typically form parts of acronyms, abbreviations, or chemical formulas.

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

of

-1.83

と

-1.50

 more

-1.48

for

-1.46

 even

-1.41

 also

-1.40

 That

-1.37

........

-1.34

to

-1.34

POSITIVE LOGITS

 aluminio

1.59

 inond

1.54

 bascul

1.46

ïc

1.45

 tornillo

1.43

 tamaños

1.41

 diminuer

1.41

 comportements

1.38

 réparer

1.37

 nadru

1.37

Activations Density 0.136%