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

HIP

-0.72

 elev

-0.71

 liner

-0.70

 retard

-0.69

 champion

-0.68

LG

-0.67

 organizing

-0.66

 champ

-0.65

 feder

-0.64

 hoops

-0.64

POSITIVE LOGITS

Cra

0.84

chem

0.81

emen

0.80

emis

0.80

yip

0.79

ebus

0.78

Nar

0.77

emonium

0.75

earth

0.75

atown

0.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.