INDEX

Explanations

response to requests and stimuli

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

prepositions and phrases indicating relationships between entities, particularly in formal communication or responses.

oai_token-act-pair · claude-3-7-sonnet-20250219 Triggered by @neilrathi

The neuron spikes on procedural/modality markers in scientific‐style method statements (e.g. “will,” “be,” “using,” “may,” “include,” “investigated,” “determined,” etc.).

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

 ayudan

-1.41

いても

-1.34

そのため

-1.34

紹

-1.29

enggarakan

-1.27

⌣

-1.26

 affectionate

-1.23

 interpersonal

-1.23

 переда

-1.21

 siete

-1.21

POSITIVE LOGITS

 responds

1.68

 Response

1.62

 response

1.62

ﹶ

1.41

໊

1.35

 Respond

1.34

 reacts

1.30

respond

1.26

 Reaction

1.25

Responses

1.25

Activations Density 0.076%

response to requests and stimuli

prepositions and phrases indicating relationships between entities, particularly in formal communication or responses.

The neuron spikes on procedural/modality markers in scientific‐style method statements (e.g. “will,” “be,” “using,” “may,” “include,” “investigated,” “determined,” etc.).

No Comments

No Known Activations

response to requests and stimuli

prepositions and phrases indicating relationships between entities, particularly in formal communication or responses.

The neuron spikes on procedural/modality markers in scientific‐style method statements (e.g. “will,” “be,” “using,” “may,” “include,” “investigated,” “determined,” etc.).

No Comments

No Known Activations