INDEX

Explanations

multiplication and tensor products

The neuron fires on tokens representing mathematical binary‐operation symbols (e.g. “times,” “otimes,” “kron,” etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 utilização

-1.38

()

-1.26

 configuração

-1.23

 plages

-1.23

 dispositifs

-1.22

肯定会

-1.21

 utiliser

-1.20

 relazione

-1.19

 incluse

-1.16

);

-1.16

POSITIVE LOGITS

⽣

1.32

传统的

1.30

to

1.25

هو

1.24

普通的

1.23

cetamol

1.22

 jätte

1.21

 guerras

1.21

 with

1.20

new

1.19

Activations Density 0.007%