INDEX

Explanations

mentions of locations, specifically cities

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

names of cities

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

Pru

-0.44

Wob

-0.41

󠁣

-0.41

runApp

-0.40

ندما

-0.40

跚

-0.40

AutoScale

-0.40

 tram

-0.40

 tapp

-0.39

yszer

-0.38

POSITIVE LOGITS

City

0.75

 City

0.73

 CITY

0.65

ValueStyle

0.63

CITY

0.61

manjaro

0.58

Casey

0.57

ྪ

0.55

StringCopy

0.55

 Cité

0.55

Activations Density 0.008%

mentions of locations, specifically cities

names of cities

No Comments

No Known Activations