INDEX

Explanations

needs immediate or nothing

The neuron detects the beginnings and key pronouns of relative clauses (e.g. “who,” “whom,” “that,” “for whom,” “given”) used to introduce distinctions or definitions.

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

 frequentemente

-1.30

ąc

-1.06

 avancer

-1.05

 tých

-1.04

 spesso

-1.02

intérieur

-1.02

 peinados

-1.02

 dôle

-1.02

tså

-1.01

 occaf

-1.01

POSITIVE LOGITS

but

2.48

 only

1.91

 without

1.48

 nhưng

1.27

 with

1.27

but

1.20

 even

1.19

as

1.16

 ONLY

1.16

 although

1.15

Activations Density 0.057%