INDEX

Explanations

that followed by a word

The neuron fires on words that introduce or signal a causal or subordinate clause—most prominently the word “that” as well as verbs like “requires” and “cause.”

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

が

-1.95

を

-1.81

などを

-1.73

<bos>

-1.73

って思

-1.69

覚悟

-1.65

だが

-1.59

 Because

-1.55

ですが

-1.52

問題は

-1.52

POSITIVE LOGITS

corrhi

2.03

を探す

1.52

んですよ

1.48

の外

1.43

のもの

1.40

 fateful

1.39

感じで

1.38

ocarditis

1.38

produzione

1.38

ſſen

1.38

Activations Density 0.073%