INDEX

Explanations

Lam Research Corporation

The neuron fires strongly on blocks that read like spammy or promotional copy packed with brand/product names, trending buzzwords, and other jargon rather than normal narrative 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

 keep

-1.59

เคย

-1.48

by

-1.43

了个

-1.41

 nifty

-1.40

 teilweise

-1.39

 gettin

-1.35

 phenomenal

-1.34

 atriz

-1.33

If

-1.32

POSITIVE LOGITS

 occasion

1.48

tiy

1.30

onta

1.30

褲

1.30

osing

1.29

ﻂ

1.28

欣賞

1.27

lowing

1.26

">“

1.25

nly

1.25

Activations Density 0.003%