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

vre

-0.97

eral

-0.88

bol

-0.84

ascript

-0.83

amo

-0.83

eki

-0.83

azon

-0.79

xual

-0.78

bp

-0.78

awed

-0.74

POSITIVE LOGITS

 theoret

0.68

 Doctrine

0.66

Topic

0.66

akedown

0.65

Tru

0.63

 doctrines

0.61

 timetable

0.60

 extradition

0.60

 Annotations

0.59

å¥

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.