INDEX

Explanations

Erik wa qq 1 IntArray jar

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

periods in URLs, version numbers, and non-English text (especially Chinese and Japanese).

oai_token-act-pair · claude-4-5-sonnet Triggered by @bfgcn7frf5

The neuron activates on numeric and dot‐containing tokens (e.g. digits, decimals, abbreviations with “.”, and sentence‐ending periods).

oai_token-act-pair · o4-mini Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

).

-2.92

-2.61

the

-2.16

-2.14

),

-2.09

An

-2.05

-2.03

when

-1.98

 garner

-1.95

POSITIVE LOGITS

🫃

2.52

廋

2.36

嫘

2.28

呲

2.23

selben

2.23

ſelves

2.20

鼴

2.20

 ,‬

2.19

⭓

2.16

腠

2.16

Activations Density 0.009%

Erik wa qq 1 IntArray jar

periods in URLs, version numbers, and non-English text (especially Chinese and Japanese).

The neuron activates on numeric and dot‐containing tokens (e.g. digits, decimals, abbreviations with “.”, and sentence‐ending periods).

No Comments

No Known Activations

Erik wa qq 1 IntArray jar

periods in URLs, version numbers, and non-English text (especially Chinese and Japanese).

The neuron activates on numeric and dot‐containing tokens (e.g. digits, decimals, abbreviations with “.”, and sentence‐ending periods).

No Comments

No Known Activations