INDEX

Explanations

various kinds

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

The neuron activates on key content words—mostly nouns and verbs carrying real semantic weight rather than on common function words.

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

the word "to" when it precedes verbs or indicates purpose/direction in formal or instructional contexts.

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

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

 niemals

0.96

 bullies

0.86

 haters

0.86

 quando

0.83

 cuando

0.81

 perros

0.75

 когда

0.74

 nunca

0.74

 when

0.73

 bukan

0.73

POSITIVE LOGITS

 বেশকিছু

1.15

 కీలక

1.01

 σημαν

0.89

いくつかの

0.87

了一些

0.86

 notamment

0.85

بعض

0.84

 některé

0.82

Some

0.81

诸多

0.81

Activations Density 0.686%

various kinds

The neuron activates on key content words—mostly nouns and verbs carrying real semantic weight rather than on common function words.

the word "to" when it precedes verbs or indicates purpose/direction in formal or instructional contexts.

No Comments

No Known Activations

various kinds

The neuron activates on key content words—mostly nouns and verbs carrying real semantic weight rather than on common function words.

the word "to" when it precedes verbs or indicates purpose/direction in formal or instructional contexts.

No Comments

No Known Activations