INDEX

Explanations

duration and time periods

The neuron is responding to unusually rare or low‐frequency words (i.e. it fires strongest on tokens that are least common across the corpus).

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

 gewähr

-1.31

 aisladas

-1.28

 financieras

-1.13

gucigalpa

-1.11

 ordinaria

-1.10

 respetar

-1.09

 Realität

-1.09

 cubiertas

-1.05

 jähr

-1.04

 Aufmerksamkeit

-1.01

POSITIVE LOGITS

for

7.50

 لمدة

3.44

 selama

2.81

 για

2.75

משך

2.45

 برای

2.44

 przez

2.34

 протягом

2.34

for

2.22

For

2.14

Activations Density 0.418%