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

artney

-0.77

agher

-0.77

akespe

-0.76

esta

-0.75

achment

-0.74

vo

-0.71

aterasu

-0.68

acts

-0.67

yrus

-0.67

vari

-0.66

POSITIVE LOGITS

TY

0.76

 safer

0.70

 nickel

0.69

 partName

0.68

pole

0.67

 Handle

0.66

 Hate

0.63

£ı

0.60

ezvous

0.60

fork

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.