INDEX

Explanations

references to racial identity and historical context related to race

oai_token-act-pair · gpt-4o-mini Triggered by @bot

asking questions using auxiliary verbs

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

the starts of sentences or clauses — especially discourse markers, modals, and stance/transition words (e.g., "So", "Would", "Will", "Maybe", "Lets").

oai_token-act-pair · gpt-5-mini Triggered by @vetterc0

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ſelves

-0.54

 dezelve

-0.54

delwed

-0.50

 ſind

-0.48

 ervan

-0.47

 yalnızca

-0.46

ckså

-0.46

 imidlertid

-0.46

 freilich

-0.44

tę

-0.44

POSITIVE LOGITS

ConstraintMaker

0.55

SourceChecksum

0.54

 ballon

0.52

 coté

0.51

周泽楷

0.50

UnknownFieldSet

0.46

 loosing

0.46

endlich

0.46

AntiForgeryToken

0.45

 betek

0.45

Activations Density 1.371%

references to racial identity and historical context related to race

asking questions using auxiliary verbs

the starts of sentences or clauses — especially discourse markers, modals, and stance/transition words (e.g., "So", "Would", "Will", "Maybe", "Lets").

No Comments

No Known Activations

references to racial identity and historical context related to race

asking questions using auxiliary verbs

the starts of sentences or clauses — especially discourse markers, modals, and stance/transition words (e.g., "So", "Would", "Will", "Maybe", "Lets").

No Comments

No Known Activations