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

adobe

-0.78

leon

-0.68

canon

-0.66

powers

-0.66

 Lilith

-0.64

opolis

-0.64

omorph

-0.63

swing

-0.63

cery

-0.63

 Elements

-0.62

POSITIVE LOGITS

 Cheong

0.73

ILLE

0.72

onte

0.71

imeo

0.70

essage

0.69

ccording

0.69

INTON

0.69

ETA

0.68

 captcha

0.67

Je

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.