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

Not in Any Lists

No Comments

Negative Logits

 Topic

-0.82

matter

-0.79

Rect

-0.74

TextColor

-0.72

 showc

-0.69

CSI

-0.66

 Nurse

-0.64

ACP

-0.63

Network

-0.63

each

-0.63

POSITIVE LOGITS

abies

0.71

 selling

0.71

iking

0.66

iven

0.64

 tram

0.64

 souven

0.64

antine

0.64

anth

0.64

haw

0.63

 monet

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.