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

bard

-0.79

giene

-0.72

 neigh

-0.67

awks

-0.66

 insign

-0.66

DL

-0.66

etooth

-0.65

aca

-0.64

ardless

-0.63

 feas

-0.63

POSITIVE LOGITS

imov

0.79

oute

0.64

 Gaddafi

0.64

 subt

0.61

 Sheen

0.60

HTML

0.60

 Melbourne

0.59

 Etsy

0.59

 Croatia

0.59

 Dempsey

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.