INDEX

Explanations

phrases associated with self-reference and identity

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

reflexive pronouns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.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

 createSlice

-0.30

 cows

-0.30

openConnection

-0.29

Xna

-0.29

Gob

-0.29

ならず

-0.29

dash

-0.29

long

-0.29

poc

-0.29

łuż

-0.28

POSITIVE LOGITS

 zichzelf

0.70

 AssemblyTitle

0.63

 themselves

0.62

 himself

0.59

 herself

0.57

GEBURTS

0.57

 الحره

0.56

 resourceCulture

0.56

 himſelf

0.56

给自己

0.54

Activations Density 0.145%

phrases associated with self-reference and identity

reflexive pronouns

No Comments

No Known Activations