INDEX

Explanations

special characters and code syntax

The neuron is broadly triggered by very frequent, generic tokens (common words and short unit markers) that recur across different contexts.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

0.77

 এলি

0.75

Gr

0.75

0.74

com

0.74

ferencia

0.73

Pon

0.72

ally

0.72

 Steps

0.72

 with

0.71

POSITIVE LOGITS

ين

0.85

Marne

0.84

theless

0.83

팥

0.81

IRA

0.78

फेसर

0.78

唥

0.78

ंदाज

0.77

HAD

0.77

Rahul

0.76

Activations Density 0.000%