INDEX

Explanations

to be

The neuron strongly activates on words that introduce predictions or expectations—i.e. future-oriented cue verbs like “expect,” “appear,” “likely,” or “be expected.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 arqué

0.73

 pretending

0.66

"/>.

0.64

ยนต์

0.63

 czyli

0.61

я

0.61

and

0.61

м

0.59

 goodbye

0.58

 enclosing

0.58

POSITIVE LOGITS

ğin

0.66

Cohort

0.64

Aile

0.61

 awọn

0.60

Cordial

0.60

Code

0.59

ರಳ

0.59

ğinde

0.59

Công

0.59

Activations Density 0.000%