INDEX

Explanations

actions and achievements

The neuron is firing on very high-frequency English words—common function words and common verbs (e.g. “to,” “see,” “shown,” “been,” “write,” “check,” “is,” etc.).

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

<bos>

-1.33

 three

-1.30

 before

-1.30

 diminish

-1.23

 after

-1.22

if

-1.21

 possibly

-1.20

These

-1.18

 placing

-1.13

 pertaining

-1.13

POSITIVE LOGITS

1.70

to

1.59

 into

1.33

quely

1.27

 ainsi

1.16

 cyclisme

1.15

ulates

1.12

lude

1.11

the

1.10

into

1.09

Activations Density 0.037%