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

Chance

-0.72

OTUS

-0.71

 epile

-0.68

 Cancer

-0.64

 weeds

-0.64

uberty

-0.64

 Chop

-0.64

 Potato

-0.63

AI

-0.62

 foes

-0.62

POSITIVE LOGITS

ibur

0.93

deen

0.90

Â·Â·

0.89

etheless

0.89

peak

0.78

rongh

0.76

ICO

0.74

perties

0.74

EngineDebug

0.73

kish

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.