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

chell

-0.76

pieces

-0.75

dn

-0.67

piece

-0.66

TING

-0.65

SPONSORED

-0.64

nik

-0.63

bish

-0.62

 escapes

-0.62

getting

-0.61

POSITIVE LOGITS

 Flavoring

0.83

 Hilbert

0.81

 Caucus

0.78

 Valhalla

0.74

 Dresden

0.72

arella

0.68

ftime

0.67

 Gillespie

0.67

 Nebula

0.66

 Corridor

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.