INDEX

Explanations

The neuron fires on hex-digit tokens (0–9, A–F) in hex dumps or other sequences of hexadecimal values.

oai_token-act-pair · o4-mini

Triggered by @jyhe0408

Oxygenation

np_acts-logits-general · gemini-2.5-flash-lite

hexadecimal numbers and digit sequences in code or technical contexts.

oai_token-act-pair · claude-3-7-sonnet-20250219

Triggered by @neilrathi

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_22/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 frites

-1.22

 that

-1.21

and

-1.20

but

-1.10

 ristorante

-1.09

 Accueil

-1.08

ttemberg

-1.05

 tricot

-1.04

 захворювання

-1.03

 mignon

-1.02

POSITIVE LOGITS

Liefer

1.30

Hersteller

1.24

including

1.15

翃

1.14

 včetně

1.13

 durante

1.10

 obter

1.09

玊

1.09

 grumpy

1.09

Sicherheit

1.07

Activations Density 0.007%

The neuron fires on hex-digit tokens (0–9, A–F) in hex dumps or other sequences of hexadecimal values.

Oxygenation

hexadecimal numbers and digit sequences in code or technical contexts.

No Known Activations

The neuron fires on hex-digit tokens (0–9, A–F) in hex dumps or other sequences of hexadecimal values.

Oxygenation

hexadecimal numbers and digit sequences in code or technical contexts.

No Known Activations