INDEX

Explanations

specific terminators for code or technical descriptions

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

The neuron fires strongly on the very first word of a new section or standalone heading.

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

the word "use" or phrases containing "to use" in instructional or descriptive contexts.

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

common standalone whole-word tokens (often simple verbs or adjectives) rather than subword fragments.

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

 estava

0.88

বিড়

0.81

ྰ

0.80

yep

0.78

 assurent

0.77

 Фі

0.77

没想到

0.76

ನೆ

0.75

ृत्व

0.74

قال

0.74

POSITIVE LOGITS

mit

0.84

0.82

0.79

lld

0.75

nt

0.74

deliver

0.74

lt

0.73

nal

0.73

receive

0.72

swap

0.72

Activations Density 0.001%

specific terminators for code or technical descriptions

The neuron fires strongly on the very first word of a new section or standalone heading.

the word "use" or phrases containing "to use" in instructional or descriptive contexts.

common standalone whole-word tokens (often simple verbs or adjectives) rather than subword fragments.

No Comments

No Known Activations