INDEX

Explanations

import aliases

The neuron strongly fires on the token “weight,” i.e. it is specifically detecting occurrences of the word “weight” (as seen in the SQL query).

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

pubescens

0.91

tand

0.85

tis

0.83

takes

0.83

GLE

0.81

 উঠিয়া

0.79

steins

0.78

види

0.78

tik

0.77

tang

0.77

POSITIVE LOGITS

inine

0.90

ymmet

0.83

ذج

0.82

 PYTHON

0.75

 bowel

0.75

 כן

0.74

糞

0.73

sembles

0.72

 ಕಂಪ

0.72

 সমালোচনা

0.72

Activations Density 0.000%