INDEX

Explanations

database query methods

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

method names in code that retrieve data by identifier or query parameter.

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

Explanation of neuron 4 behavior: the main thing this neuron does is find data-access method calls beginning with “find” (e.g. findById, findByEmail, findAll).

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

 iNdEx

-1.46

brtc

-1.38

woners

-1.38

приятие

-1.37

 Infór

-1.36

 ferien

-1.31

 geplaatst

-1.31

 stol

-1.31

ruf

-1.30

内の

-1.29

POSITIVE LOGITS

Other

1.77

There

1.48

The

1.44

1.42

Using

1.42

1.38

 häufigsten

1.38

Perhaps

1.37

Despite

1.35

 Other

1.34

Activations Density 0.005%

database query methods

method names in code that retrieve data by identifier or query parameter.

Explanation of neuron 4 behavior: the main thing this neuron does is find data-access method calls beginning with “find” (e.g. findById, findByEmail, findAll).

No Comments

No Known Activations

database query methods

method names in code that retrieve data by identifier or query parameter.

Explanation of neuron 4 behavior: the main thing this neuron does is find data-access method calls beginning with “find” (e.g. findById, findByEmail, findAll).

No Comments

No Known Activations