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

pres

-0.88

unct

-0.85

uncture

-0.83

picture

-0.77

yl

-0.76

izable

-0.75

clud

-0.72

uration

-0.71

urally

-0.71

ãĥĺãĥ©

-0.70

POSITIVE LOGITS

 Typhoon

0.80

sbm

0.79

 Daly

0.78

 Jihad

0.76

 Kham

0.69

 Daughter

0.68

 Schneider

0.68

Bei

0.66

 Mayer

0.66

 Nolan

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.