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

DragonMagazine

-0.97

-0.92

pse

-0.79

MpServer

-0.76

culosis

-0.75

 pestic

-0.73

utenberg

-0.72

oola

-0.71

 Cosponsors

-0.71

axter

-0.70

POSITIVE LOGITS

OP

0.68

ICT

0.63

ise

0.62

elle

0.62

 Merry

0.62

orth

0.61

lat

0.61

vict

0.60

 hail

0.60

Í

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.