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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

advertisement

-0.71

 Subtle

-0.71

versions

-0.69

ibble

-0.68

 amen

-0.66

 Atmosp

-0.66

Dial

-0.63

umbnails

-0.62

 âĨĳ

-0.62

interstitial

-0.62

POSITIVE LOGITS

 Sole

0.73

bilt

0.72

Val

0.70

worst

0.70

shire

0.69

 Summer

0.67

cum

0.67

rit

0.66

ourn

0.66

 Ange

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.