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

Negative Logits

song

-0.76

ritz

-0.72

 Seah

-0.70

llo

-0.67

Song

-0.66

 Ashes

-0.65

 Frey

-0.64

EH

-0.63

iosyncr

-0.63

sers

-0.62

POSITIVE LOGITS

é¾įå¥ĳå£«

0.76

 trusts

0.69

tyr

0.69

 citizenship

0.69

 guessed

0.66

icago

0.65

izon

0.64

 curls

0.63

Osw

0.63

database

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.