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

ciating

-0.97

HCR

-0.84

assetsadobe

-0.82

nir

-0.82

enegger

-0.81

CLASSIFIED

-0.81

è¦ļéĨĴ

-0.80

brainer

-0.78

)=(

-0.78

 cumbers

-0.76

POSITIVE LOGITS

 Mason

0.76

 Grade

0.74

 Miracle

0.73

 Correctional

0.71

 Methodist

0.70

 Deposit

0.65

 Presbyterian

0.64

 Blind

0.64

 Bake

0.63

 parole

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.