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

ictionary

-0.84

GET

-0.76

TON

-0.67

byn

-0.67

ãĥīãĥ©

-0.66

ById

-0.66

Pak

-0.64

CRIP

-0.63

Anonymous

-0.63

LOAD

-0.62

POSITIVE LOGITS

 Flavoring

0.75

ller

0.74

 Dealer

0.71

 Shuttle

0.68

ãĥ¼ãĥĨãĤ£

0.67

 Nanto

0.67

 Trooper

0.67

 Firm

0.65

 Memor

0.64

Hon

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.