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

IZ

-0.78

BY

-0.68

ãĥİ

-0.67

Subject

-0.65

Islamic

-0.64

Thom

-0.63

åĽ

-0.62

ommod

-0.61

Author

-0.61

fram

-0.61

POSITIVE LOGITS

naire

0.83

ï¸

0.77

 Spotlight

0.73

Au

0.72

 Monstrous

0.68

 Appl

0.67

yrus

0.66

 Huck

0.65

ologue

0.65

naires

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.