INDEX

Explanations

when starting previously

The neuron fires on subordinating and adversative cue words (e.g. if, when, but, then) that introduce conditional or causal clauses.

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

(!__

-1.43

we

-1.42

 that

-1.38

牍

-1.35

水泥

-1.34

である

-1.30

 they

-1.29

-1.27

лення

-1.27

POSITIVE LOGITS

就會

1.44

reiches

1.31

攝

1.27

Ṭ

1.25

 auffi

1.23

 یا

1.23

 núcleos

1.23

tų

1.21

으면

1.21

 torres

1.20

Activations Density 0.162%