INDEX

Explanations

communication results derived upper bounded

The neuron is detecting the progression through a long technical passage—it steadily increases activation for tokens that occur later (i.e. further along) in the abstract or document.

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

achute

-0.91

liyor

-0.84

اهرة

-0.82

 reprezent

-0.82

 skateboard

-0.81

]._

-0.81

EIP

-0.78

дии

-0.76

 wifi

-0.75

jvm

-0.75

POSITIVE LOGITS

triangleq

1.46

≜

1.28

 IEEE

1.09

 WINNER

1.06

IEEE

1.05

 receiver

1.02

 fading

0.96

 receivers

0.93

 impairments

0.93

SNR

0.91

Activations Density 0.068%