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

rocket

-0.68

termin

-0.67

 rehears

-0.64

bleacher

-0.62

itte

-0.62

backer

-0.61

thumbnails

-0.60

 signings

-0.60

soDeliveryDate

-0.60

arching

-0.58

POSITIVE LOGITS

icity

0.70

 Theft

0.69

estone

0.68

emis

0.67

agan

0.67

iary

0.67

cery

0.67

ores

0.65

since

0.63

 since

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.