INDEX

Explanations

sequential connectors and conjunctions

This neuron flags tokens that occur in dialogue or script‐style excerpts—i.e. speaker names and words in quoted spoken lines (including contractions).

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

 Drž

-2.13

ſu

-2.05

recently

-2.05

 rése

-1.91

ครื่อง

-1.91

⎔

-1.91

 Você

-1.89

prettier

-1.88

€”

-1.85

媗

-1.82

POSITIVE LOGITS

 with

2.25

and

2.23

2.22

 before

2.17

 when

2.11

 during

2.08

or

1.92

1.91

or

1.89

 then

1.88

Activations Density 0.010%