INDEX

Explanations

body parts and positions

The neuron strongly activates on words naming bodily actions or positions (e.g. lunge, guard, parry, fists, arm-in-arm, extended).

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

 acidente

-1.47



-1.39

 décisions

-1.34

 ingenuity

-1.34

 inov

-1.33

OIL

-1.32

PTT

-1.32

 here

-1.31

 singola

-1.31



-1.30

POSITIVE LOGITS

\_

1.46

里面

1.43

1.42

to

1.41

 piernas

1.39

1.34

Have

1.33

1.31

 referencias

1.30

 fronteras

1.30

Activations Density 0.008%