INDEX

Explanations

documentation parameter types

The neuron activates on parameter‐documentation annotations (e.g. “@param” tags) in code comments.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 interessante

-2.17

 insanely

-2.09

 unbelievably

-2.02

 interwoven

-2.00

 masterful

-1.99

 ravaged

-1.96

 coloridos

-1.95

 intertwined

-1.94

 overarching

-1.93

 hermosos

-1.92

POSITIVE LOGITS

 Some

2.34

 selecting

2.30

 There

2.25

 taking

2.19

鹛

2.16

 Although

2.09

 stating

2.03

 before

2.00

 Depending

1.99

on

1.98

Activations Density 0.001%