INDEX

Explanations

north east corner or system race

The neuron consistently lights up on BPE subword fragments of specialized or uncommon terms (e.g. parts of proper nouns, technical jargon, or domain‐specific words).

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

 these

-1.84

of

-1.66

-1.52

the

-1.49

 both

-1.43

 time

-1.41

jangkau

-1.37

这两个

-1.31

这两

-1.30

 эти

-1.27

POSITIVE LOGITS

着一个

1.54

䜣

1.51

ѕ

1.48

𖣔

1.43

έρα

1.41

༛

1.41

EFIAPI

1.38

</h2>

1.38

着一

1.37

DispatchToProps

1.37

Activations Density 0.203%