INDEX

Explanations

years and dates

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

The neuron is picking up on Wikipedia‐style structural tokens—especially the “TOC”/“NO TOC” magic words and standalone page‐number markers.

oai_token-act-pair · o4-mini

Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_34/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 lược

-0.80

实验室

-0.77

的一声

-0.75

 windshield

-0.72

攻撃

-0.70

壱

-0.70

 intolerant

-0.67

 laboratories

-0.66

zda

-0.66

 graciosas

-0.66

POSITIVE LOGITS

 tecnológico

0.68

 complied

0.67

ITC

0.67

contemporary

0.65

 układ

0.65

åt

0.65

 geest

0.64

 contemporáneo

0.64

prada

0.63

Celsius

0.62

Activations Density 0.103%

years and dates

The neuron is picking up on Wikipedia‐style structural tokens—especially the “TOC”/“NO TOC” magic words and standalone page‐number markers.

No Known Activations

years and dates

The neuron is picking up on Wikipedia‐style structural tokens—especially the “TOC”/“NO TOC” magic words and standalone page‐number markers.

No Known Activations

years and dates

The neuron is picking up on Wikipedia‐style structural tokens—especially the “__TOC__”/“__NO TOC__” magic words and standalone page‐number markers.

No Known Activations

years and dates

The neuron is picking up on Wikipedia‐style structural tokens—especially the “__TOC__”/“__NO TOC__” magic words and standalone page‐number markers.

No Known Activations

The neuron is picking up on Wikipedia‐style structural tokens—especially the “TOC”/“NO TOC” magic words and standalone page‐number markers.

The neuron is picking up on Wikipedia‐style structural tokens—especially the “TOC”/“NO TOC” magic words and standalone page‐number markers.