INDEX

Explanations

New York City train problems

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

The neuron lights up on tokens that carry concrete factual details—things like names, dates, addresses, measurements, or other specific “hard” information.

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

the word "if" or phrases introducing conditional statements.

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

common function words and other structural markers in text, often appearing in addresses, dates, or URLs.

oai_token-act-pair · gpt-5 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

ações

0.81

 Gild

0.80

 глава

0.79

 pequeña

0.78

 какой

0.78

 desarrollado

0.78

 algumas

0.75

ea

0.75

 dezelfde

0.75

ação

0.73

POSITIVE LOGITS

申し

0.80

ہر

0.80

بر

0.79

تا

0.77

бре

0.77

HOSTNAME

0.76

فر

0.75

خ

0.74

モニ

0.73

ک

0.72

Activations Density 0.000%

New York City train problems

The neuron lights up on tokens that carry concrete factual details—things like names, dates, addresses, measurements, or other specific “hard” information.

the word "if" or phrases introducing conditional statements.

common function words and other structural markers in text, often appearing in addresses, dates, or URLs.

No Comments

No Known Activations