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

arten

-0.99

rek

-0.83

ega

-0.75

ummer

-0.72

anchester

-0.71

orship

-0.71

aye

-0.71

Ã¨re

-0.70

riter

-0.69

agar

-0.69

POSITIVE LOGITS

Kings

0.72

 Suns

0.68

 DRAG

0.68

Bull

0.67

 Dragons

0.66

bats

0.65

humans

0.65

ONS

0.64

depending

0.63

 Prism

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.