INDEX

Explanations

sentence endings followed by new clauses

The neuron fires on blank‐line or paragraph‐break tokens, i.e. it detects boundaries between sections of the text.

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

前言

-1.30

仙台

-1.18

适用

-1.08

徳島

-1.07

湘南

-1.06

岡山

-1.06

睾

-1.06

函館

-1.05

āj

-1.04

為

-1.04

POSITIVE LOGITS

but

1.91

 then

1.45

yet

1.27

 verschillende

1.27

但是

1.20

を始め

1.20

に行って

1.14

が好き

1.09

に行く

1.07

 podrían

1.02

Activations Density 0.108%