INDEX

Explanations

location or following description

The neuron fires most strongly on tokens that signal “the following” or “below” (for example, Japanese/Chinese 以下).

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

of

-2.77

of

-2.09

-2.03

 остальные

-2.02

就是

-2.00

)$

-1.91

to

-1.90

ную

-1.89

และ

-1.88

,"

-1.85

POSITIVE LOGITS



2.19

Однако

2.03

नलोड

1.93

 reformas

1.92

 Saltar

1.90

 thump

1.84

lüğü

1.82

 sociedades

1.78

1.77

﹍﹍

1.77

Activations Density 0.002%