INDEX

Explanations

Liviu Negoiță, Cost effectiveness, Q

The neuron strongly activates on tokens conveying quantitative data—e.g. numbers, dates, measurements, and other numeric information.

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

レーキ

-0.86

 patin

-0.84

 tami

-0.84

 piste

-0.84

 Chama

-0.82

 snowboard

-0.79

麓

-0.79

 reparti

-0.77

 families

-0.77

 Chiara

-0.77

POSITIVE LOGITS

 swim

1.95

 swimming

1.70

 swimmers

1.59

 Swim

1.53

 swims

1.48

Swim

1.43

 swimmer

1.32

 Swimming

1.31

Swimming

1.25

swim

1.24

Activations Density 0.012%