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

cember

-0.82

EStream

-0.80

tails

-0.77

uthor

-0.77

hammad

-0.75

elson

-0.75

hower

-0.73

mouth

-0.73

deen

-0.72

bledon

-0.70

POSITIVE LOGITS

 stru

0.78

-+-+

0.68

 Fargo

0.68

 sovere

0.66

essor

0.64

clus

0.64

 Etsy

0.64

Occup

0.62

 Constantin

0.62

 Adin

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.