INDEX

Explanations

works fine

The neuron strongly activates on very common, high-frequency English words (e.g. “company,” “on,” “works,” “fine,” “often,” “look,” “down,” etc.).

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

 that

-1.45

 mikina

-1.23

 répondit

-1.20

 détruit

-1.19

獁

-1.16

(^_^

-1.13

―――

-1.13

 Dette

-1.13

っしゃ

-1.13

POSITIVE LOGITS

is

1.53

1.16

 قطاع

1.15

なども

1.13

 операций

1.06

 kuchen

1.05

 Dateien

1.03

 الشرطة

1.00

),

0.99

前は

0.99

Activations Density 0.085%