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

iability

-0.70

ARC

-0.68

uez

-0.66

 Saber

-0.66

CTR

-0.62

ILLE

-0.62

 partName

-0.62

iyah

-0.61

 Emacs

-0.60

kB

-0.60

POSITIVE LOGITS

htaking

0.76

 nomine

0.68

terness

0.67

hower

0.67

eeper

0.67

 vain

0.66

 proble

0.64

pite

0.64

hift

0.64

liest

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.