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

ovember

-0.78

jay

-0.73

ogy

-0.72

orks

-0.72

 Gleaming

-0.71

neum

-0.71

aceae

-0.71

DragonMagazine

-0.70

©¶æ

-0.70

alus

-0.69

POSITIVE LOGITS

 Views

0.66

Typ

0.66

Ultra

0.64

 Readers

0.63

 Aerial

0.63

DIV

0.63

 Venezuel

0.63

 Reader

0.63

 Sphere

0.62

 Think

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.