INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

utterstock

-0.81

 heels

-0.77

Newsletter

-0.71

FactoryReloaded

-0.68

 mirror

-0.67

 gloss

-0.66

 cloak

-0.66

 insign

-0.66

cyclopedia

-0.65

 pretext

-0.64

POSITIVE LOGITS

\\\\\\\\

0.78

ares

0.73

ggie

0.71

prising

0.71

querque

0.69

LOS

0.68

cott

0.67

tes

0.67

olls

0.65

Pwr

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.