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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rehens

-0.71

Veh

-0.69

 grunt

-0.68

 accomplishment

-0.67

 millisec

-0.66

 unbeat

-0.64

entry

-0.64

 childbirth

-0.64

ienne

-0.63

llular

-0.62

POSITIVE LOGITS

udeau

0.72

Ã°

0.70

ä¹ĭ

0.66

CVE

0.66

çīĪ

0.63

soType

0.62

APS

0.61

oil

0.60

 Moran

0.60

KH

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.