INDEX

Explanations

still then

The neuron fires on numeric tokens and quantitative expressions (e.g. integers and decimals).

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

Allgemeine

-0.84

 reliably

-0.74

预期

-0.74

 ไม่

-0.74

举例

-0.73

minente

-0.73

 рестайлинг

-0.73

いちご

-0.73

 Abuja

-0.72

 Inicial

-0.72

POSITIVE LOGITS

 still

1.02

ülé

1.00

 Entonces

0.98

 allora

0.96

still

0.93

 then

0.92

แล้ว

0.87

 alors

0.86

 revealing

0.85

Still

0.84

Activations Density 0.068%