INDEX

Explanations

after specific punctuations or words

The neuron responds to low‐frequency, content‐bearing tokens (i.e. uncommon or specialized words) in the text.

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

ሮች

0.87

вол

0.87

 Acest

0.86

 зустрі

0.85

 lavoratori

0.84

 સફેદ

0.82

Ω

0.82

 ಬಾರಿ

0.81

ድግዳ

0.81

Ал

0.80

POSITIVE LOGITS

0.76

ht

0.75

0.71

dy

0.71

 operations

0.70

NG

0.67

ones

0.66

de

0.65

 finishing

0.65

apsible

0.65

Activations Density 0.000%