INDEX

Explanations

phrases indicating authorship or contributions to a research paper

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

Competing interests

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

 passe

-0.38

how

-0.35

 Oppen

-0.32

stateProvider

-0.32

 pass

-0.31

olism

-0.30

try

-0.30

 into

-0.30

 solu

-0.29

posX

-0.28

POSITIVE LOGITS

isContained

0.79

IsMutable

0.77

 EconPapers

0.74

GOTREF

0.67

WithIOException

0.65

DockStyle

0.65

VersionUID

0.65

GeoNames

0.63

PerformLayout

0.63

migrationBuilder

0.63

Activations Density 0.006%

phrases indicating authorship or contributions to a research paper

Competing interests

No Comments

No Known Activations