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

iPhone

-0.86

iak

-0.73

-+

-0.68

 Mazda

-0.67

dayName

-0.64

 wholesale

-0.62

mouth

-0.61

heng

-0.60

toggle

-0.59

icone

-0.59

POSITIVE LOGITS

doms

0.76

dom

0.70

 Ivory

0.68

Assembly

0.67

ugu

0.66

kson

0.66

 Bone

0.64

VEN

0.63

 bombard

0.63

 Laur

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.