INDEX

Explanations

personal names

This neuron detects four-digit numeric tokens (particularly years) in the text.

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

-1.25

All

-1.24

or

-1.23

and

-1.23

The

-1.22

-1.19

At

-1.19

 However

-1.17

-1.16

POSITIVE LOGITS

↵↵↵↵

1.71

↵↵↵↵↵↵↵

1.66

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.65

↵↵↵↵↵↵↵↵↵↵↵↵↵

1.62

↵↵↵↵↵↵↵↵

1.62

↵↵↵↵↵

1.59

<eos>

1.58

↵↵↵↵↵↵↵↵↵↵↵

1.55

↵↵↵

1.55

↵

1.55

Activations Density 0.020%