INDEX

Explanations

results presented

The neuron detects references to specific figure panels or parts (e.g. “Panel B,” “top left,” “presented in” etc.) in scientific or technical documents.

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

 конструкции

-1.45

};

-1.45

考えた

-1.44

-1.42

 frühen

-1.41

 которого

-1.41

 damaligen

-1.36

 систему

-1.34

 którego

-1.29

}+

-1.28

POSITIVE LOGITS

毉

1.74

all

1.55

 haji

1.52

our

1.47

 wali

1.46

 results

1.43

capulco

1.41

㍑

1.41

 alip

1.41

🗿

1.40

Activations Density 0.004%