INDEX

Explanations

references to chapters and sections within a text

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

chapter and section identifiers

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

 réessayer

-0.42

 للمعارف

-0.34

<eos>

-0.34

 opérateur

-0.32

 használ

-0.32

ceğine

-0.31

 AssemblyCompany

-0.30

 Alembic

-0.30

complexContent

-0.30

iterranée

-0.30

POSITIVE LOGITS

 chapter

0.70

 Chapter

0.69

chapter

0.63

Chapter

0.62

챕

0.60

Chapitre

0.58

 tabPage

0.58

CHAPTER

0.57

 zwiſchen

0.57

DockStyle

0.57

Activations Density 0.021%

references to chapters and sections within a text

chapter and section identifiers

No Comments

No Known Activations