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

oca

-0.65

gem

-0.63

aith

-0.63

 lumin

-0.62

warts

-0.62

 counterparts

-0.61

 mighty

-0.60

 brethren

-0.58

 Apple

-0.58

 hitherto

-0.57

POSITIVE LOGITS

DragonMagazine

0.75

hester

0.71

prus

0.68

soDeliveryDate

0.68

 diner

0.68

ãĥ¼ãĥĨãĤ£

0.67

 tiss

0.66

ä»£

0.66

zanne

0.66

Posted

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.