INDEX

Explanations

choose the best option

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

The neuron strongly activates on words and forms of the act of choosing or deciding (e.g. “choose,” “chose,” “choosing,” “decide”).

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

New Auto-Interp

Configuration

google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ból

1.28

1.26

 khoảng

1.16

1.14

్యూ

1.13

le

1.12

 Noronha

1.11

š

1.11

ة

1.09

oš

1.08

POSITIVE LOGITS

െടുത്ത

1.85

 wisely

1.68

 lựa

1.52

 выбира

1.47

 choice

1.38

 Choice

1.37

 escolha

1.33

สรร

1.33

 выборе

1.30

 choix

1.29

Activations Density 0.135%

choose the best option

The neuron strongly activates on words and forms of the act of choosing or deciding (e.g. “choose,” “chose,” “choosing,” “decide”).

No Comments

No Known Activations

choose the best option

The neuron strongly activates on words and forms of the act of choosing or deciding (e.g. “choose,” “chose,” “choosing,” “decide”).

No Comments

No Known Activations

**choose the best option**

The neuron strongly activates on words and forms of the act of choosing or deciding (e.g. “choose,” “chose,” “choosing,” “decide”).

No Comments

No Known Activations

**choose the best option**

The neuron strongly activates on words and forms of the act of choosing or deciding (e.g. “choose,” “chose,” “choosing,” “decide”).

No Comments

No Known Activations

choose the best option

choose the best option