INDEX

Explanations

opening things

The neuron fires on words and word‐parts that indicate opening or unboxing actions (e.g. “open,” “opening,” “opened,” “unbox,” 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

 would

-1.51

was

-1.46

 with

-1.46

</h4>

-1.43

↵↵↵

-1.29

er

-1.28

不太

-1.23

-1.21

to

-1.21

 wouldn

-1.20

POSITIVE LOGITS

 unopened

1.41

unun

1.40

 emballage

1.37

 dibuka

1.35

ممنون

1.35

狴

1.34

lossen

1.34

话题

1.34

 opened

1.32

خدا

1.31

Activations Density 0.037%