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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pic

-0.73

 removing

-0.69

 tolerated

-0.67

 Woodward

-0.63

 tram

-0.62

McA

-0.62

fav

-0.62

 disreg

-0.61

 cleaners

-0.60

 Foot

-0.59

POSITIVE LOGITS

UGE

0.79

halla

0.75

Brow

0.75

Ã±o

0.75

hran

0.72

urized

0.71

phia

0.71

gewater

0.70

urat

0.68

idency

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.