INDEX

Explanations

body parts and their adjacent areas

This neuron fires on imperative directional and positional terms (e.g. pulling, holding, moving body parts up/down/behind/forward) in exercise‐style instructions.

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

 ludzi

0.64

etc

0.62

等的

0.61

したり

0.61

 इत्यादि

0.61

 тощо

0.60

等

0.59

等等

0.59

などの

0.59

等の

0.59

POSITIVE LOGITS

 near

0.73

 abdomen

0.71

 where

0.70

abdomen

0.70

 thorax

0.67

 fossa

0.67

 flange

0.66

 torso

0.65

 pectoral

0.65

 slopes

0.64

Activations Density 0.370%