INDEX

Explanations

The neuron activates strongly on degree‐of‐comparison and modality phrases—e.g. “as popular as,” modals like “will,” and adverbs such as “never” or “always.”

oai_token-act-pair · o4-mini

Triggered by @jyhe0408

modal verbs (especially "will") expressing future possibility or prediction.

oai_token-act-pair · claude-4-5-sonnet

Triggered by @jyhe0408

uses of modal auxiliaries that express intention, prediction, possibility, or obligation.

oai_token-act-pair · gpt-5

Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

savefig

0.54

숨

0.52

दर्श

0.51

StringBuilder

0.51

后面的

0.51

われている

0.50

较低

0.48

较高

0.48

已被

0.47

గు

0.47

POSITIVE LOGITS

 never

4.59

 никогда

4.34

 nunca

4.22

 NEVER

4.22

never

4.19

 nigdy

4.06

 Never

4.00

Never

3.88

 jamais

3.80

 nooit

3.77

Activations Density 1.068%

The neuron activates strongly on degree‐of‐comparison and modality phrases—e.g. “as popular as,” modals like “will,” and adverbs such as “never” or “always.”

modal verbs (especially "will") expressing future possibility or prediction.

uses of modal auxiliaries that express intention, prediction, possibility, or obligation.

No Known Activations

The neuron activates strongly on degree‐of‐comparison and modality phrases—e.g. “as popular as,” modals like “will,” and adverbs such as “never” or “always.”

modal verbs (especially "will") expressing future possibility or prediction.

uses of modal auxiliaries that express intention, prediction, possibility, or obligation.

No Known Activations