INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

iott

-0.88

kefeller

-0.80

ensical

-0.78

ocamp

-0.76

orld

-0.75

ledge

-0.73

uca

-0.73

quarters

-0.72

oric

-0.72

ean

-0.70

POSITIVE LOGITS

âĿ

0.81

 ILCS

0.80

Mek

0.74

Contents

0.72

ULE

0.72

 Shap

0.68

Islamic

0.66

FIL

0.66

Fil

0.65

Kop

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.