INDEX

Explanations

output release

The main thing this neuron does is pick out cooking‐style recipe instructions, especially numbered steps, quantities, and related action verbs.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 原子

0.56

 Atomic

0.55

原子

0.55

 Supplier

0.55

 Suppliers

0.54

 Atom

0.54

 Stiftung

0.54

 austen

0.53

 match

0.52

 Atoms

0.52

POSITIVE LOGITS

输出

1.25

 output

1.19

 输出

1.17

output

1.15

 outputs

1.09

輸出

1.09

Output

1.05

 Output

1.05

Outputs

1.02

OUTPUT

1.02

Activations Density 0.451%