INDEX

Explanations

Output

This neuron activates on tokens that represent numeric information or identifiers—e.g. numbers, spelled-out numbers, file/line references, error codes, or hex-style IDs.

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

設け

0.75

যারা

0.72

 предусматри

0.68

用い

0.65

designer

0.65

パートナー

0.65

 परमा

0.64

WebApi

0.61

partnership

0.61

DESIGN

0.60

POSITIVE LOGITS

ırmızı

0.84

 但是

0.82

 angrily

0.80

 noticeably

0.76

 输出

0.76

 segundos

0.74

 suspiciously

0.74

 hello

0.73

 ValueError

0.73

 Interestingly

0.72

Activations Density 0.047%