INDEX

Explanations

references to geographic locations and neighborhoods

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

neighborhood and area names

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

رشف

-0.51

лтамалар

-0.42

 Numerade

-0.40

httphttps

-0.40

Välislingid

-0.40

Diwedd

-0.39

nalités

-0.37

publicain

-0.37

minLength

-0.37

Vanjske

-0.35

POSITIVE LOGITS

 neighborhoods

0.75

 suburb

0.65

 neighborhood

0.65

 área

0.62

 area

0.59

 neighbourhoods

0.58

neighborhood

0.57

 areas

0.56

 áreas

0.54

 suburbs

0.53

Activations Density 0.030%

references to geographic locations and neighborhoods

neighborhood and area names

No Comments

No Known Activations