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

 largeDownload

-0.70

×Ĳ

-0.67

 Revel

-0.66

 Canaver

-0.66

 jurisdiction

-0.65

 Lomb

-0.65

 Asuka

-0.64

 asylum

-0.63

EStream

-0.63

================================================================

-0.62

POSITIVE LOGITS

yl

1.38

idy

0.83

ym

0.83

ylene

0.83

zanne

0.80

ole

0.78

eph

0.78

ummies

0.77

inks

0.75

ivery

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.