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

 Guinness

-0.79

EngineDebug

-0.75

 Gutenberg

-0.75

aceae

-0.75

acebook

-0.74

ilater

-0.72

âĹ¼

-0.69

Math

-0.69

ography

-0.68

ographies

-0.66

POSITIVE LOGITS

agar

0.79

quartered

0.76

lay

0.75

anger

0.69

aido

0.69

 exting

0.67

ook

0.67

 ende

0.67

ugi

0.66

abor

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.