INDEX

Explanations

relative clauses

The neuron chiefly responds to numeric tokens—that is, it fires on numbers.

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

and

-1.53

 then

-1.48

 this

-1.43

it

-1.28

 different

-1.23

revista

-1.20

Suka

-1.16

 соответству

-1.16

 some

-1.13

 each

-1.13

POSITIVE LOGITS

 which

4.53

who

4.22

 whose

2.89

 الذي

2.75

 которая

2.61

 که

2.45

 που

2.45

 التي

2.39

whose

2.36

ที่

2.34

Activations Density 0.059%