INDEX

Explanations

numerical values and formatting related to research metrics or citations

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

academic papers and equations

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

ⓧ

-0.56

Portale

-0.53

béco

-0.52

saan

-0.50

AnchorTagHelper

-0.50

imageshack

-0.50

ticiones

-0.49

つて

-0.47

Aktualisiert

-0.47

chevêque

-0.46

POSITIVE LOGITS

 contigo

0.59

 Ganzen

0.53

 StatefulWidget

0.53

 متعلقه

0.52

bibinfo

0.50

épaisseur

0.50

perma

0.50

 hereafter

0.50

 Playhouse

0.50

 multitasking

0.50

Activations Density 0.047%

numerical values and formatting related to research metrics or citations

academic papers and equations

No Comments

No Known Activations