INDEX

Explanations

citations and references

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

This neuron detects numeric reference or citation identifiers (numbers in parentheses) in scholarly text.

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

There’s insufficient evidence to infer its feature, as it did not activate on any tokens in these passages.

oai_token-act-pair · gpt-5 Triggered by @jyhe0408

years in the 2000s within academic citations and formal document metadata.

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

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_22/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ouvertes

-1.19

 téléph

-1.18

 ladr

-1.16

おそ

-1.14

 informaci

-1.13

事も

-1.12

corrhi

-1.12

 tôt

-1.12

 ondu

-1.12

and

-1.11

POSITIVE LOGITS

但却

1.20

1.19

1.09

↵

1.01

 first

0.98

is

0.97

--

0.97

などが

0.95

おそらく

0.94

 while

0.93

Activations Density 0.009%

citations and references

This neuron detects numeric reference or citation identifiers (numbers in parentheses) in scholarly text.

There’s insufficient evidence to infer its feature, as it did not activate on any tokens in these passages.

years in the 2000s within academic citations and formal document metadata.

No Comments

No Known Activations

citations and references

This neuron detects numeric reference or citation identifiers (numbers in parentheses) in scholarly text.

There’s insufficient evidence to infer its feature, as it did not activate on any tokens in these passages.

years in the 2000s within academic citations and formal document metadata.

No Comments

No Known Activations