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

uploads

-0.78

 horm

-0.70

\/\/

-0.70

 comprom

-0.69

enegger

-0.65

terness

-0.65

imov

-0.62

 unpop

-0.61

 infringing

-0.61

sembly

-0.60

POSITIVE LOGITS

tips

0.73

Ã§a

0.68

allic

0.67

tip

0.67

berus

0.66

 Orient

0.66

igmatic

0.66

chrom

0.64

onica

0.64

ately

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.