INDEX

Explanations

steady-state line case

This neuron flags technical terms used as labels for formal constructs (e.g. geometric “line,” steady‐“state,” worst‐case complexity) that typically head definitions or key concepts.

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

却被

-2.14

 SEVENT

-2.08

却没有

-2.00

Hæ

-2.00

 sélections

-1.97

 “(

-1.90

 Atención

-1.88

 نيوز

-1.84

但却

-1.84

POSITIVE LOGITS

for

2.30

2.16

л

2.13

呦

1.88

So

1.88

with

1.88

 certainly

1.85

of

1.84

...

1.82

1.80

Activations Density 0.007%