INDEX

Explanations

technical and biographical terms

This neuron fires on words that appear as the very first token of a block or line—i.e. section-heading or line-opening tokens (like “System”, “released”, “#define”, “using”, “rights,” etc.).

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

乄

-1.30

-1.20

鷥

-1.20

鲩

-1.16

齟

-1.15

cji

-1.09

ريقة

-1.08

岨

-1.08

也不敢

-1.06

 inneces

-1.06

POSITIVE LOGITS

Both

1.25

Just

1.13

Give

1.13

 both

1.09

one

1.08

 eleven

1.06

Often

1.05

two

1.05

Several

1.02

</em>

1.02

Activations Density 0.000%