INDEX

Explanations

results and implications

The neuron highlights high‐information “result” or “consequence” verbs and phrases (e.g. allows, suggests, provides, points to, forces, associated with) that signal findings or causal links in a report.

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

 this

-2.00

it

-1.44

 where

-1.23

as

-1.20

its

-1.19

 after

-1.16

 when

-1.15

 these

-1.14

 GALERÍA

-1.11

 their

-1.10

POSITIVE LOGITS

همچنین

1.29

她也

1.29

1.25

 also

1.22

他也

1.21

 confirmó

1.20

1.19

 proporcionan

1.19

 reveló

1.19

Activations Density 0.085%