INDEX

Explanations

connecting to context

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

transitional or connective phrases that introduce explanations, comparisons, or methods (such as "to", "by", "as", "especially in", "such as").

oai_token-act-pair · claude-4-5-sonnet Triggered by @bfgcn7frf5

The neuron fires most strongly on common high‐frequency function words (prepositions, conjunctions, articles, etc.) and similar little “stop” tokens.

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

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/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

 isolada

-1.45

 потому

-1.45

viembre

-1.38

"]}

-1.37

 protože

-1.35

 trám

-1.34

 BECAUSE

-1.34

牖

-1.34

 same

-1.33

罴

-1.32

POSITIVE LOGITS

Some

1.55

or

1.48

 ....

1.46

1.44

 this

1.41

 Some

1.41

1.38

 .....

1.37

.....

1.33

 HUGE

1.33

Activations Density 0.134%

connecting to context

transitional or connective phrases that introduce explanations, comparisons, or methods (such as "to", "by", "as", "especially in", "such as").

The neuron fires most strongly on common high‐frequency function words (prepositions, conjunctions, articles, etc.) and similar little “stop” tokens.

No Comments

No Known Activations

connecting to context

transitional or connective phrases that introduce explanations, comparisons, or methods (such as "to", "by", "as", "especially in", "such as").

The neuron fires most strongly on common high‐frequency function words (prepositions, conjunctions, articles, etc.) and similar little “stop” tokens.

No Comments

No Known Activations