INDEX

Explanations

we found or they found

The neuron fires on author‐reporting phrases—especially the pronouns “we” or “they” combined with the auxiliary “have” that introduce reported results.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 kepentingan

-1.14

 diskusi

-1.13

()}

-1.07

 immagine

-1.04

 importancia

-1.03

 kön

-1.02

?\\

-1.02

بسم

-1.01

。。。。

-1.01

 keputusan

-1.00

POSITIVE LOGITS

 thấy

1.73

 found

1.55

พบ

1.45

find

1.20

found

1.19

 দেখ

1.19

 finding

1.12

saw

1.11

 обнаружи

1.11

 findings

1.08

Activations Density 0.088%