INDEX

Explanations

numerical or coding patterns related to document references or citations

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

URLs and file paths

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

LookAnd

-0.71

ImageContext

-0.65

RenderAtEndOf

-0.63

.*")]

-0.61

 crossorigin

-0.61

GEBURTS

-0.60

AndEndTag

-0.60

ſicht

-0.59

HasForeignKey

-0.59

 FetchType

-0.57

POSITIVE LOGITS

 simplifié

0.32

 ujednoznacz

0.31

 hjemmeside

0.30

 PeEnEo

0.30

AFFIRMED

0.30

 disambiguazione

0.28

 Genehmigung

0.28

üyada

0.27

afficheront

0.27

 Tentang

0.26

Activations Density 0.039%

numerical or coding patterns related to document references or citations

URLs and file paths

No Comments

No Known Activations