INDEX

Explanations

drawing comparisons

The neuron fires on words and short phrases that set up analogies or comparisons (e.g. “similar to,” “much like,” “akin to,” “thought of as”).

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

囑

-1.21

anym

-1.20

 uchun

-1.13

-1.12

 ayur

-1.11

 byen

-1.09

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.07

⁶

-1.05

ཝ

-1.05

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.05

POSITIVE LOGITS

 like

2.13

2.08

but

2.03

 would

1.62

an

1.61

 from

1.45

 similar

1.41

if

1.39

 using

1.36

so

1.23

Activations Density 0.045%