INDEX

Explanations

Japanese characters and specific words

The neuron fires on function words that introduce subordinate or relative clauses—e.g. “to,” “that,” “where.”

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

 Maybe

-1.26

 that

-1.20

toDouble

-1.19

 modalidades

-1.18

 Several

-1.18

 Some

-1.17

 Actually

-1.16

 Obviously

-1.15

 Would

-1.13

 fábricas

-1.13

POSITIVE LOGITS

を

1.41

 …..

1.29

 .......

1.28

 .........

1.27

 poë

1.27

の

1.26

 ........

1.24

は

1.23

が

1.22

 capilla

1.21

Activations Density 0.039%