INDEX

Explanations

programming commands and literals

The neuron fires on code tokens that embed numeric values (e.g. enum‐style constant names or parameters with underscores and attached digits).

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

and

-2.03

all

-1.71

 with

-1.55

에도

-1.55

 również

-1.49

 became

-1.46

 becomes

-1.45

。”

-1.45

?’

-1.44

for

-1.43

POSITIVE LOGITS

1.62

 hebben

1.51

 weten

1.43

ጴ

1.38

 женщины

1.37

鱨

1.37

 ſta

1.36

が少ない

1.34

jika

1.34

[{\

1.34

Activations Density 0.030%