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

yrinth

-0.76

ao

-0.69

die

-0.66

icit

-0.66

ombs

-0.65

aries

-0.62

united

-0.62

meet

-0.61

comings

-0.61

faith

-0.60

POSITIVE LOGITS

avorite

0.76

laus

0.70

amac

0.69

×©

0.67

 Carney

0.67

kef

0.67

×ĳ

0.66

ovember

0.65

assium

0.65

DragonMagazine

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.