variable names ending in s

map to int conversion (negative activations)

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 256 tokens each

Dataset (Dashboard)

neuronpedia/python-code-simplified

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ject

-0.16

 number

-0.13

 current

-0.10

one

-0.09

ome

-0.07

ase

-0.06

_grid

-0.05

0.00

POSITIVE LOGITS

tices

0.34

times

0.29

typ

0.17

ays

0.17

its

0.15

bers

0.15

 found

0.14

cles

0.13

lay

0.12

Activations Density 100.000%