INDEX

Explanations

method or means

The neuron fires on content nouns—single-word entity or object names (e.g. “drones,” “career,” “image,” “population,” “controller,” etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

(!__

-1.31

に変更

-1.20

為に

-1.16

を作成

-1.10

rightarrow

-1.10

 eenig

-1.09

Into

-1.03

IBAction

-1.03

OSPITAL

-1.02

に入れる

-1.02

POSITIVE LOGITS

by

3.77

 with

2.88

 using

2.72

 through

2.67

via

2.20

in

1.62

ด้วย

1.59

 mediante

1.58

 bằng

1.58

 pomocí

1.55

Activations Density 0.217%