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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Blueprint

-0.72

gaard

-0.69

naire

-0.69

 Swanson

-0.65

lund

-0.63

soType

-0.63

 Shutterstock

-0.62

Î²

-0.60

ulator

-0.58

Sv

-0.58

POSITIVE LOGITS

me

1.05

us

0.80

ciating

0.78

 them

0.78

pport

0.76

plom

0.71

 hers

0.70

perture

0.68

0.67

FontSize

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.