INDEX

Explanations

Q: A:

The neuron activates on code‐style identifier tokens—like JSON object keys or function/local names—within structured data or code snippets.

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

 their

-1.34

 After

-1.27

 used

-1.27

 With

-1.23

It

-1.19

 Ultimately

-1.17

can

-1.17

 Some

-1.16

 which

-1.16

get

-1.11

POSITIVE LOGITS

nalité

1.16

 colossal

1.14

 incrí

1.13

粜

1.09

still

1.08

鵙

1.07

 choisis

1.07

ふわふわ

1.05

 vários

1.03

素晴らしい

1.03

Activations Density 0.000%