INDEX

Explanations

actions followed by "together" or "away"

The neuron responds to dynamic action verbs—especially less common or vivid verbs indicating an activity or change.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 সম্পূর্ণরূপে

0.66

 Slight

0.60

 Conce

0.57

 linken

0.57

 Properly

0.57

 Fully

0.56

 prinsip

0.56

 grac

0.55

 ments

0.55

 Selv

0.55

POSITIVE LOGITS

 away

1.13

out

0.97

away

0.84

Away

0.84

 AWAY

0.77

 Away

0.77

 aside

0.71

出来的

0.70

 together

0.67

出一

0.66

Activations Density 0.476%