INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

club

-0.77

ocal

-0.73

combe

-0.73

cop

-0.71

RPG

-0.71

 Grounds

-0.71

aghetti

-0.71

haus

-0.70

hare

-0.68

ront

-0.67

POSITIVE LOGITS

merce

0.78

 wrench

0.71

 diagn

0.66

tions

0.64

 characterization

0.64

 bulbs

0.63

nih

0.63

BALL

0.63

Hew

0.62

 bulb

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.