INDEX

Explanations

affirmation or negation

This neuron strongly activates on numeric tokens (including digits and decimal points) and similar non-alphabetic answer tokens.

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

 have

-2.41

 där

-2.22

<bos>

-2.02

 thrived

-1.97

-1.89

使える

-1.88

おすすめの

-1.84

前は

-1.77

 waar

-1.75

晄

-1.75

POSITIVE LOGITS

狴

2.77

кога

2.31

 adalah

2.27

沚

2.17

stanti

2.14



2.14

 sabemos

2.08

🫃

2.08

 باشند

2.06

がたくさん

2.05

Activations Density 0.002%