INDEX

Explanations

German shepherd and German names

The neuron activates on occurrences of the token “German” (or its subtoken “german”) wherever it appears, whether referring to the nationality or as part of a word like “germanium.”

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

您好

-0.88

 illegitimate

-0.85

quíni

-0.84

 opener

-0.83

Доб

-0.83

Ere

-0.82

BC

-0.82

改善

-0.82

意義

-0.81

你好

-0.81

POSITIVE LOGITS

 Shepherd

1.34

Shepherd

1.03

otta

0.93

 Shepard

0.91

 Shepherds

0.90

教练

0.89

icus

0.86

loyd

0.85

engineered

0.84

很想

0.84

Activations Density 0.030%