INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ortium

-0.74

aterasu

-0.71

antage

-0.70

 Buckingham

-0.69

 sworn

-0.68

 contradicted

-0.64

soType

-0.64

ointed

-0.64

azeera

-0.63

ational

-0.63

POSITIVE LOGITS

killer

0.73

WARE

0.73

Topic

0.71

 Volunte

0.70

Ghost

0.68

Hand

0.68

Vo

0.66

Folder

0.66

Writer

0.66

DragonMagazine

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.