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

SourceFile

-0.75

atal

-0.70

Kar

-0.69

then

-0.66

ando

-0.65

ãĤ¬

-0.65

humans

-0.64

ocol

-0.64

NetMessage

-0.64

Pakistan

-0.64

POSITIVE LOGITS

byter

0.67

 impe

0.65

hew

0.65

 referen

0.65

ellectual

0.65

 nerd

0.64

 tongue

0.64

 grade

0.64

afort

0.64

esides

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.