INDEX

Explanations

if statements and definitions

This neuron detects the key “proof‐directive” or statement‐introducing words in math text (e.g. Let, Suppose, Prove, If, Show, Then).

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

躾

-1.42

</h4>

-1.40

 он

-1.32

⎆

-1.30

ẠI

-1.27

獰

-1.27

 incred

-1.26

⃠

-1.25

 diventato

-1.25

giày

-1.25

POSITIVE LOGITS

cisive

1.58

Normally

1.46

肌肤

1.34

illou

1.34

This

1.30

 една

1.30

substack

1.29

 台中

1.25

 Wählen

1.24

才行

1.24

Activations Density 0.032%