INDEX

Explanations

negation or negative concepts

The neuron consistently fires on occurrences of “negate,” “negating,” “negation,” or the uppercase “NEG,” i.e. tokens conveying the root “negat” (negation).

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

躥

-2.61

翃

-2.39

 INFORMACIÓN

-2.38

-2.22



-2.22

}))

-2.16

𝔗

-2.14

增添

-2.13

épendant

-2.11

derdag

-2.09

POSITIVE LOGITS

↵

4.34

咴

2.05

 stadig

1.88

寕

1.84

 diría

1.81

 devenus

1.80

 hablamos

1.78

赑

1.77

‌های

1.72

 estaría

1.69

Activations Density 0.003%