INDEX

Explanations

specific names and titles, particularly in the context of brands, organizations, or significant figures

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

Chameleon, Empowered, stereotactic radiosurgery

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

 EconPapers

-0.40

 respec

-0.39

Trayectoria

-0.34

プリング

-0.34

 jurisdic

-0.33

 themselves

-0.31

(!__

-0.30

Kre

-0.30

insertBefore

-0.29

est

-0.29

POSITIVE LOGITS

 חיצוניים

0.68

 Normdatei

0.67

__*/

0.66

 HasFactory

0.63

MessageOf

0.60

 שוליים

0.60

πως

0.54

 Exacts

0.54

copg

0.53

Бележки

0.53

Activations Density 0.231%

specific names and titles, particularly in the context of brands, organizations, or significant figures

Chameleon, Empowered, stereotactic radiosurgery

No Comments

No Known Activations