INDEX

Explanations

The neuron fires on the very first word of each sentence (i.e. sentence‐initial tokens).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 adalah

0.90

ですが

0.90

です

0.83

，

0.77

だが

0.74

。

0.70

虽然

0.69

雖然

0.68

でしたが

0.67

 है

0.66

POSITIVE LOGITS

 they

1.64

it

1.56

we

1.55

 there

1.55

 они

1.35

 anyone

1.30

 anybody

1.30

you

1.23

she

1.23

he

1.21

Activations Density 0.409%