INDEX

Explanations

ending punctuation

The neuron primarily detects numeric expressions and quantitative figures (e.g. counts, measurements, percentages).

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

 only

-1.50

を

-1.39

そのため

-1.27

が

-1.20

參考文獻

-1.19

もしか

-1.19

 steadily

-1.15

も

-1.13

<bos>

-1.13

風味

-1.13

POSITIVE LOGITS

の準備

1.33

corrhi

1.32

って何

1.31

の方に

1.30

に来

1.29

 zelve

1.25

 accesorio

1.20

 aguj

1.20

性は

1.20

書の

1.20

Activations Density 0.134%