INDEX

Explanations

repeated references to locations or places

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

the word here

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

ObservableList

-0.30

 německ

-0.30

 vlastní

-0.30

 giacca

-0.30

of

-0.30

 batata

-0.30

Mutable

-0.30

 Given

-0.29

 disponibilità

-0.29

 given

-0.29

POSITIVE LOGITS

richTextPanel

0.76

Geplaatst

0.67

 فريبيس

0.65

 للاسماء

0.61

tagHelperRunner

0.60

 وتسجيلات

0.59

WireFormatLite

0.59

AndEndTag

0.58

躇

0.57

üyor

0.57

Activations Density 0.013%

repeated references to locations or places

the word here

No Comments

No Known Activations