INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Jackets

-0.80

cart

-0.72

liter

-0.71

 Rooms

-0.68

 Hours

-0.67

 Online

-0.67

KEN

-0.66

onson

-0.64

 Clever

-0.63

 Heights

-0.62

POSITIVE LOGITS

idth

0.79

DragonMagazine

0.78

Reviewer

0.72

yk

0.72

 Cosponsors

0.65

ym

0.63

bleacher

0.62

¬¼

0.62

istan

0.62

phies

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.