INDEX

Explanations

only if

The neuron fires almost exclusively on the token “to.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ifiably

1.69

ians

1.19

1.08

UATION

1.03

ራሉ

1.03

ग्विजय

1.02

(".",

1.01

iid

1.00

 plainly

0.99

 ionized

0.99

POSITIVE LOGITS

о

1.67

ва

1.61

ка

1.52

у

1.49

و

1.46

 второй

1.40

так

1.35

ти

1.31

дца

1.31

ра

1.30

Activations Density 0.044%