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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 perfected

-0.70

 Trin

-0.67

 GOODMAN

-0.63

 converted

-0.63

eds

-0.62

 Reader

-0.61

 pioneered

-0.60

 constructed

-0.60

hift

-0.59

 equivalents

-0.59

POSITIVE LOGITS

ļéĨĴ

0.87

agall

0.87

 tremend

0.83

æ©

0.81

agraph

0.81

allery

0.77

ifax

0.76

akedown

0.74

riage

0.73

wcs

0.73

Activations Density 0.000%

No Known Activations

This feature has no known activations.