INDEX

Explanations

and explanations

The neuron is responding to very high‐frequency function words (common English stop words such as “if,” “the,” “is,” “what,” “when,” etc.).

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

 with

0.63

0.60

and

0.58

for

0.58

 from

0.57

 which

0.55

0.54

 before

0.53

if

0.52

POSITIVE LOGITS

 SalesLT

0.81

 GoObject

0.74

 Сы

0.68

 Jaaye

0.67

uradaki

0.67

 Quadrupèdes

0.67

 Теннис

0.66

 Oiseaux

0.64

 素材

0.64

iftoire

0.64

Activations Density 0.503%