INDEX

Explanations

technical documents and acronyms

This neuron fires on tokens that are repeated consecutively (i.e. runs of the same token).

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

 from

-1.46

 There

-1.31

of

-1.31

 Because

-1.30

 While

-1.29

 From

-1.27

 Using

-1.26

 over

-1.25

 What

-1.25

or

-1.24

POSITIVE LOGITS

などを

1.25

 résultats

1.15

 ..........

1.09

もしか

1.08

紧接着

1.07

干脆

1.05

 ........

1.05

 CONTIN

1.04

就可以了

1.04

 deterior

1.02

Activations Density 0.007%