INDEX

Explanations

receiving prestigious awards

The neuron is highly activated by numeric tokens—especially standalone digits and multi‐digit year values—i.e. it detects occurrences of numbers and dates.

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

 دسترسی

-0.72

一八

-0.70

 best

-0.70

 most

-0.69

_{*}

-0.69

 Reson

-0.68

 hanggang

-0.67

৫

-0.67

 ferien

-0.67

 effekt

-0.67

POSITIVE LOGITS

 Presidential

0.87

plots

0.82

 awarded

0.82

Receive

0.81

receive

0.81

pting

0.77

 Mukherjee

0.77

 erhielt

0.76

Plot

0.75

 distro

0.72

Activations Density 0.027%