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

 Dism

-0.75

 Moran

-0.70

 Communism

-0.69

bie

-0.69

wik

-0.68

 Occupy

-0.67

 Cecil

-0.65

Corp

-0.65

MLA

-0.63

ments

-0.63

POSITIVE LOGITS

aunders

0.78

ouble

0.76

qv

0.76

olphins

0.74

avid

0.73

oul

0.71

iamond

0.70

al

0.69

aq

0.69

alist

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.