INDEX

Explanations

acid

The neuron activates strongly for the term "acid" and related chemical/scientific terminology in scientific texts.

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

 sorta

-0.96

 сих

-0.92

migrationBuilder

-0.88

樹脂

-0.88

来年

-0.81

你说

-0.79

 advocates

-0.78

 соло

-0.77

 administ

-0.77

 honestly

-0.77

POSITIVE LOGITS

acid

1.13

ulated

1.05

Acid

1.04

 rain

1.02

 reflux

1.01

ulous

1.00

 acid

0.98

ophilus

0.97

 احترام

0.95

 corrosive

0.94

Activations Density 0.037%